diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/README.md b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..d4576fe074287232d3836bf69c21d3f2593290d9
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/README.md
@@ -0,0 +1,9 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+
+- PEFT 0.4.0
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/adapter_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..45e27418190fb0e63b4b544dca57bdb25908f0f4
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/adapter_config.json
@@ -0,0 +1,26 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "/data/wiedmann/hub/models--lmms-lab--llava-onevision-qwen2-7b-ov/snapshots/0b07bf7565e244cf4f39982249eafe8cd799d6dd",
+  "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",
+    "v_proj",
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "down_proj",
+    "gate_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}
\ No newline at end of file
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/adapter_model.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..689b90529342545ac6f958fe0d6e802716d7ebee
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:739d441fd1b78c7af6dd767f97e6ca80c375bb9ac2d3e5673f149ab9ae36d5f7
+size 692127130
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..d9989c3ebebe651b269c3202d5362ec1fbe695a5
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/config.json
@@ -0,0 +1,68 @@
+{
+  "_name_or_path": "/data/wiedmann/hub/models--lmms-lab--llava-onevision-qwen2-7b-ov/snapshots/0b07bf7565e244cf4f39982249eafe8cd799d6dd",
+  "add_faster_video": false,
+  "add_time_instruction": false,
+  "architectures": [
+    "LlavaQwenForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "faster_token_stride": 10,
+  "force_sample": false,
+  "hidden_act": "silu",
+  "hidden_size": 3584,
+  "image_aspect_ratio": "square",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": null,
+  "image_split_resolution": null,
+  "image_token_index": 151646,
+  "initializer_range": 0.02,
+  "intermediate_size": 18944,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "mm_hidden_size": 1152,
+  "mm_newline_position": "no_token",
+  "mm_patch_merge_type": "spatial_unpad",
+  "mm_projector_lr": null,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": "streaming_agg",
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_streaming_frames_per_chunk": 4,
+  "mm_streaming_input_dim": 1152,
+  "mm_streaming_mlp_ratio": 4.0,
+  "mm_streaming_num_heads": 8,
+  "mm_streaming_num_layers": 4,
+  "mm_streaming_num_state_tokens": 4096,
+  "mm_streaming_patches_per_frame": 729,
+  "mm_streaming_state_dim": 1152,
+  "mm_streaming_vision_chunk_size": 4,
+  "mm_tunable_parts": "mm_mlp_adapter,mm_vision_resampler",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "google/siglip-so400m-patch14-384",
+  "mm_vision_tower_lr": null,
+  "model_type": "llava_qwen",
+  "num_attention_heads": 28,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 4,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": 131072,
+  "tie_word_embeddings": false,
+  "tokenizer_model_max_length": 8192,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.40.0.dev0",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "use_sliding_window": false,
+  "vision_tower_pretrained": null,
+  "vocab_size": 152064
+}
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/generation_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..19a297221acb87418d4388a3decef2282c6d7316
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/generation_config.json
@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.05,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.40.0.dev0"
+}
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/non_lora_trainables.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..4d180a103d8bbf652c50a83577e678e8b58e7e31
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf4b7702cd4d5ad4950906ee87b27fd2f5fd43122024e5dc453b457447ba7084
+size 216160380
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/trainer_state.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..c0a10967625d9873764fb5f3a4f4d01eb7aacca4
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_final2_lora_agg_fc4_s4096/trainer_state.json
@@ -0,0 +1,14968 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9992893521260215,
+  "eval_steps": 100,
+  "global_step": 2110,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.85153865814209,
+      "learning_rate": 1.5625e-06,
+      "loss": 1.3426,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.559672355651855,
+      "learning_rate": 3.125e-06,
+      "loss": 1.4059,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 16.074081420898438,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 1.338,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.054110527038574,
+      "learning_rate": 6.25e-06,
+      "loss": 1.2648,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.242310523986816,
+      "learning_rate": 7.8125e-06,
+      "loss": 1.3495,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.123887062072754,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 1.1269,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.728862762451172,
+      "learning_rate": 1.09375e-05,
+      "loss": 0.7611,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.069932460784912,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7502,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.8440940380096436,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 0.6299,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.8849802017211914,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.5753,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5396977663040161,
+      "learning_rate": 1.71875e-05,
+      "loss": 0.4603,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2284892797470093,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.4925,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9219491481781006,
+      "learning_rate": 2.0312500000000002e-05,
+      "loss": 0.3113,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9894838333129883,
+      "learning_rate": 2.1875e-05,
+      "loss": 0.4576,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8990053534507751,
+      "learning_rate": 2.34375e-05,
+      "loss": 0.4276,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7087737321853638,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3926,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6728258728981018,
+      "learning_rate": 2.6562500000000002e-05,
+      "loss": 0.3082,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5877570509910583,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 0.3879,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5740751028060913,
+      "learning_rate": 2.96875e-05,
+      "loss": 0.3273,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6738312840461731,
+      "learning_rate": 3.125e-05,
+      "loss": 0.3107,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5729997754096985,
+      "learning_rate": 3.2812500000000005e-05,
+      "loss": 0.2859,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.639545738697052,
+      "learning_rate": 3.4375e-05,
+      "loss": 0.3701,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5942142605781555,
+      "learning_rate": 3.59375e-05,
+      "loss": 0.3127,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5892656445503235,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.4444,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5677840113639832,
+      "learning_rate": 3.90625e-05,
+      "loss": 0.2837,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6989716291427612,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.4245,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6140813827514648,
+      "learning_rate": 4.21875e-05,
+      "loss": 0.3569,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6196805238723755,
+      "learning_rate": 4.375e-05,
+      "loss": 0.2912,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5955633521080017,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 0.3531,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44241219758987427,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.2647,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5007181167602539,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 0.271,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48961982131004333,
+      "learning_rate": 5e-05,
+      "loss": 0.3375,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.658515453338623,
+      "learning_rate": 5.15625e-05,
+      "loss": 0.4088,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6141238808631897,
+      "learning_rate": 5.3125000000000004e-05,
+      "loss": 0.422,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6391209959983826,
+      "learning_rate": 5.46875e-05,
+      "loss": 0.4081,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.561816930770874,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 0.3091,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5445853471755981,
+      "learning_rate": 5.78125e-05,
+      "loss": 0.365,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6157516837120056,
+      "learning_rate": 5.9375e-05,
+      "loss": 0.4365,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5300709009170532,
+      "learning_rate": 6.0937500000000004e-05,
+      "loss": 0.3487,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44161155819892883,
+      "learning_rate": 6.25e-05,
+      "loss": 0.2732,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5040037631988525,
+      "learning_rate": 6.40625e-05,
+      "loss": 0.338,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5648464560508728,
+      "learning_rate": 6.562500000000001e-05,
+      "loss": 0.3292,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4822116196155548,
+      "learning_rate": 6.71875e-05,
+      "loss": 0.3101,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5597658157348633,
+      "learning_rate": 6.875e-05,
+      "loss": 0.3966,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.609082043170929,
+      "learning_rate": 7.031250000000001e-05,
+      "loss": 0.3993,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5004106760025024,
+      "learning_rate": 7.1875e-05,
+      "loss": 0.2335,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4758702516555786,
+      "learning_rate": 7.34375e-05,
+      "loss": 0.2872,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5550029277801514,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.4123,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5178859233856201,
+      "learning_rate": 7.65625e-05,
+      "loss": 0.3791,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6593338251113892,
+      "learning_rate": 7.8125e-05,
+      "loss": 0.332,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4983413815498352,
+      "learning_rate": 7.96875e-05,
+      "loss": 0.3075,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5851902961730957,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 0.3837,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6229225993156433,
+      "learning_rate": 8.28125e-05,
+      "loss": 0.4491,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5451405048370361,
+      "learning_rate": 8.4375e-05,
+      "loss": 0.3872,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5343197584152222,
+      "learning_rate": 8.593750000000001e-05,
+      "loss": 0.3562,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.47701945900917053,
+      "learning_rate": 8.75e-05,
+      "loss": 0.3613,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.501800537109375,
+      "learning_rate": 8.90625e-05,
+      "loss": 0.335,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.404818594455719,
+      "learning_rate": 9.062500000000001e-05,
+      "loss": 0.258,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4215477406978607,
+      "learning_rate": 9.21875e-05,
+      "loss": 0.2956,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5009216070175171,
+      "learning_rate": 9.375e-05,
+      "loss": 0.3486,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4160255193710327,
+      "learning_rate": 9.53125e-05,
+      "loss": 0.2797,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4765574336051941,
+      "learning_rate": 9.687500000000001e-05,
+      "loss": 0.3468,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4560016691684723,
+      "learning_rate": 9.84375e-05,
+      "loss": 0.3066,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5457615852355957,
+      "learning_rate": 0.0001,
+      "loss": 0.3968,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.514738917350769,
+      "learning_rate": 9.999994105751931e-05,
+      "loss": 0.3861,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.512539267539978,
+      "learning_rate": 9.999976423021617e-05,
+      "loss": 0.3311,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5410533547401428,
+      "learning_rate": 9.999946951850749e-05,
+      "loss": 0.3419,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4873526394367218,
+      "learning_rate": 9.999905692308813e-05,
+      "loss": 0.303,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.43660932779312134,
+      "learning_rate": 9.999852644493086e-05,
+      "loss": 0.3563,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.523208737373352,
+      "learning_rate": 9.999787808528638e-05,
+      "loss": 0.3015,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.49769333004951477,
+      "learning_rate": 9.999711184568333e-05,
+      "loss": 0.4603,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5623626112937927,
+      "learning_rate": 9.999622772792828e-05,
+      "loss": 0.4167,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4502825438976288,
+      "learning_rate": 9.99952257341057e-05,
+      "loss": 0.2811,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4873923659324646,
+      "learning_rate": 9.9994105866578e-05,
+      "loss": 0.3853,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4742412865161896,
+      "learning_rate": 9.99928681279855e-05,
+      "loss": 0.3416,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6207187175750732,
+      "learning_rate": 9.999151252124639e-05,
+      "loss": 0.3614,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4910104274749756,
+      "learning_rate": 9.99900390495568e-05,
+      "loss": 0.3507,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.44163739681243896,
+      "learning_rate": 9.998844771639073e-05,
+      "loss": 0.3017,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.39861860871315,
+      "learning_rate": 9.998673852550007e-05,
+      "loss": 0.2475,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5080884099006653,
+      "learning_rate": 9.998491148091457e-05,
+      "loss": 0.4038,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4914018213748932,
+      "learning_rate": 9.998296658694185e-05,
+      "loss": 0.4168,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.456467866897583,
+      "learning_rate": 9.99809038481674e-05,
+      "loss": 0.3364,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48166564106941223,
+      "learning_rate": 9.997872326945452e-05,
+      "loss": 0.394,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.42488086223602295,
+      "learning_rate": 9.997642485594436e-05,
+      "loss": 0.2386,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.37301647663116455,
+      "learning_rate": 9.997400861305591e-05,
+      "loss": 0.2677,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.46371519565582275,
+      "learning_rate": 9.99714745464859e-05,
+      "loss": 0.3265,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4532930552959442,
+      "learning_rate": 9.996882266220895e-05,
+      "loss": 0.3085,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.44449886679649353,
+      "learning_rate": 9.996605296647737e-05,
+      "loss": 0.3567,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4864048659801483,
+      "learning_rate": 9.996316546582128e-05,
+      "loss": 0.3663,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5717203617095947,
+      "learning_rate": 9.996016016704854e-05,
+      "loss": 0.4678,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4058225750923157,
+      "learning_rate": 9.995703707724474e-05,
+      "loss": 0.2736,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.48806998133659363,
+      "learning_rate": 9.995379620377318e-05,
+      "loss": 0.3713,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4165661036968231,
+      "learning_rate": 9.995043755427487e-05,
+      "loss": 0.3561,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.47120019793510437,
+      "learning_rate": 9.99469611366685e-05,
+      "loss": 0.3178,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.482258677482605,
+      "learning_rate": 9.99433669591504e-05,
+      "loss": 0.3966,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4851691424846649,
+      "learning_rate": 9.993965503019457e-05,
+      "loss": 0.4099,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.42257824540138245,
+      "learning_rate": 9.993582535855263e-05,
+      "loss": 0.3175,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4051867723464966,
+      "learning_rate": 9.99318779532538e-05,
+      "loss": 0.3437,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.42721447348594666,
+      "learning_rate": 9.992781282360486e-05,
+      "loss": 0.2573,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.484856516122818,
+      "learning_rate": 9.992362997919016e-05,
+      "loss": 0.3124,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.32508981227874756,
+      "eval_runtime": 199.4166,
+      "eval_samples_per_second": 1.71,
+      "eval_steps_per_second": 0.216,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4741864800453186,
+      "learning_rate": 9.991932942987158e-05,
+      "loss": 0.3739,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3499874472618103,
+      "learning_rate": 9.991491118578856e-05,
+      "loss": 0.1952,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5252445936203003,
+      "learning_rate": 9.991037525735794e-05,
+      "loss": 0.4488,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4262740910053253,
+      "learning_rate": 9.990572165527413e-05,
+      "loss": 0.3006,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5048425197601318,
+      "learning_rate": 9.990095039050886e-05,
+      "loss": 0.4355,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4851915240287781,
+      "learning_rate": 9.98960614743114e-05,
+      "loss": 0.3661,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.42701730132102966,
+      "learning_rate": 9.989105491820831e-05,
+      "loss": 0.2774,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44719696044921875,
+      "learning_rate": 9.988593073400354e-05,
+      "loss": 0.2713,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.46118542551994324,
+      "learning_rate": 9.98806889337784e-05,
+      "loss": 0.2959,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4939645528793335,
+      "learning_rate": 9.987532952989145e-05,
+      "loss": 0.3464,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4528026282787323,
+      "learning_rate": 9.986985253497858e-05,
+      "loss": 0.263,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.37571606040000916,
+      "learning_rate": 9.986425796195287e-05,
+      "loss": 0.2302,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.47263747453689575,
+      "learning_rate": 9.985854582400465e-05,
+      "loss": 0.3704,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4957786798477173,
+      "learning_rate": 9.985271613460144e-05,
+      "loss": 0.323,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5005558133125305,
+      "learning_rate": 9.984676890748787e-05,
+      "loss": 0.432,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5009768605232239,
+      "learning_rate": 9.984070415668574e-05,
+      "loss": 0.4067,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.48683542013168335,
+      "learning_rate": 9.983452189649387e-05,
+      "loss": 0.4575,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4298285245895386,
+      "learning_rate": 9.982822214148821e-05,
+      "loss": 0.3285,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4597395360469818,
+      "learning_rate": 9.982180490652165e-05,
+      "loss": 0.3253,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.48632317781448364,
+      "learning_rate": 9.981527020672413e-05,
+      "loss": 0.3769,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.48157766461372375,
+      "learning_rate": 9.98086180575025e-05,
+      "loss": 0.3702,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42433908581733704,
+      "learning_rate": 9.980184847454052e-05,
+      "loss": 0.3152,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4345506429672241,
+      "learning_rate": 9.979496147379883e-05,
+      "loss": 0.3448,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.41072341799736023,
+      "learning_rate": 9.978795707151492e-05,
+      "loss": 0.3386,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42164260149002075,
+      "learning_rate": 9.978083528420303e-05,
+      "loss": 0.306,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43312758207321167,
+      "learning_rate": 9.977359612865423e-05,
+      "loss": 0.2923,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44852885603904724,
+      "learning_rate": 9.976623962193626e-05,
+      "loss": 0.2881,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.46767958998680115,
+      "learning_rate": 9.975876578139355e-05,
+      "loss": 0.3558,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4880558252334595,
+      "learning_rate": 9.975117462464715e-05,
+      "loss": 0.3772,
+      "step": 129
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4066006541252136,
+      "learning_rate": 9.974346616959476e-05,
+      "loss": 0.3452,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43521568179130554,
+      "learning_rate": 9.973564043441057e-05,
+      "loss": 0.3222,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5290797352790833,
+      "learning_rate": 9.972769743754531e-05,
+      "loss": 0.3946,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.48268359899520874,
+      "learning_rate": 9.971963719772621e-05,
+      "loss": 0.3245,
+      "step": 133
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.44581907987594604,
+      "learning_rate": 9.971145973395684e-05,
+      "loss": 0.338,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5416569709777832,
+      "learning_rate": 9.970316506551725e-05,
+      "loss": 0.3702,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.44114506244659424,
+      "learning_rate": 9.969475321196373e-05,
+      "loss": 0.3113,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3807583749294281,
+      "learning_rate": 9.968622419312894e-05,
+      "loss": 0.2146,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.40282970666885376,
+      "learning_rate": 9.96775780291217e-05,
+      "loss": 0.2559,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.48782384395599365,
+      "learning_rate": 9.96688147403271e-05,
+      "loss": 0.3658,
+      "step": 139
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5037343502044678,
+      "learning_rate": 9.965993434740633e-05,
+      "loss": 0.3701,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.39732131361961365,
+      "learning_rate": 9.965093687129668e-05,
+      "loss": 0.1867,
+      "step": 141
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4830625057220459,
+      "learning_rate": 9.96418223332115e-05,
+      "loss": 0.3737,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.48340654373168945,
+      "learning_rate": 9.963259075464011e-05,
+      "loss": 0.3488,
+      "step": 143
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5103211402893066,
+      "learning_rate": 9.962324215734782e-05,
+      "loss": 0.3572,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4740847051143646,
+      "learning_rate": 9.961377656337579e-05,
+      "loss": 0.3572,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4335394501686096,
+      "learning_rate": 9.960419399504106e-05,
+      "loss": 0.3236,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5602415204048157,
+      "learning_rate": 9.959449447493643e-05,
+      "loss": 0.4375,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.37148797512054443,
+      "learning_rate": 9.958467802593046e-05,
+      "loss": 0.226,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.43919458985328674,
+      "learning_rate": 9.957474467116739e-05,
+      "loss": 0.3057,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4056735038757324,
+      "learning_rate": 9.956469443406706e-05,
+      "loss": 0.2706,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3951229155063629,
+      "learning_rate": 9.955452733832493e-05,
+      "loss": 0.2979,
+      "step": 151
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.46784666180610657,
+      "learning_rate": 9.954424340791196e-05,
+      "loss": 0.3827,
+      "step": 152
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5241360664367676,
+      "learning_rate": 9.953384266707453e-05,
+      "loss": 0.4221,
+      "step": 153
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.551415205001831,
+      "learning_rate": 9.952332514033447e-05,
+      "loss": 0.4983,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4503234624862671,
+      "learning_rate": 9.951269085248898e-05,
+      "loss": 0.3563,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.35928985476493835,
+      "learning_rate": 9.950193982861047e-05,
+      "loss": 0.2177,
+      "step": 156
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4604697823524475,
+      "learning_rate": 9.949107209404665e-05,
+      "loss": 0.3469,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.47664588689804077,
+      "learning_rate": 9.948008767442034e-05,
+      "loss": 0.331,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5417811870574951,
+      "learning_rate": 9.946898659562951e-05,
+      "loss": 0.4792,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4464086592197418,
+      "learning_rate": 9.94577688838472e-05,
+      "loss": 0.3236,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4469507038593292,
+      "learning_rate": 9.944643456552133e-05,
+      "loss": 0.311,
+      "step": 161
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.42212817072868347,
+      "learning_rate": 9.943498366737487e-05,
+      "loss": 0.2676,
+      "step": 162
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.43942126631736755,
+      "learning_rate": 9.942341621640558e-05,
+      "loss": 0.2845,
+      "step": 163
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4137977659702301,
+      "learning_rate": 9.941173223988602e-05,
+      "loss": 0.2685,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4725792109966278,
+      "learning_rate": 9.93999317653635e-05,
+      "loss": 0.303,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5099173784255981,
+      "learning_rate": 9.938801482065998e-05,
+      "loss": 0.3943,
+      "step": 166
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.48603492975234985,
+      "learning_rate": 9.937598143387205e-05,
+      "loss": 0.3565,
+      "step": 167
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.44320836663246155,
+      "learning_rate": 9.93638316333708e-05,
+      "loss": 0.2516,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.46852850914001465,
+      "learning_rate": 9.935156544780183e-05,
+      "loss": 0.3911,
+      "step": 169
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4605495035648346,
+      "learning_rate": 9.933918290608509e-05,
+      "loss": 0.3302,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.41779473423957825,
+      "learning_rate": 9.932668403741488e-05,
+      "loss": 0.3123,
+      "step": 171
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3556387424468994,
+      "learning_rate": 9.93140688712598e-05,
+      "loss": 0.2228,
+      "step": 172
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3777831494808197,
+      "learning_rate": 9.93013374373626e-05,
+      "loss": 0.245,
+      "step": 173
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4771595299243927,
+      "learning_rate": 9.928848976574019e-05,
+      "loss": 0.2657,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5206533074378967,
+      "learning_rate": 9.927552588668351e-05,
+      "loss": 0.3885,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.40793871879577637,
+      "learning_rate": 9.926244583075748e-05,
+      "loss": 0.2896,
+      "step": 176
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4514513611793518,
+      "learning_rate": 9.924924962880093e-05,
+      "loss": 0.3457,
+      "step": 177
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3950751721858978,
+      "learning_rate": 9.923593731192655e-05,
+      "loss": 0.2616,
+      "step": 178
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42353376746177673,
+      "learning_rate": 9.922250891152077e-05,
+      "loss": 0.3356,
+      "step": 179
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4029572904109955,
+      "learning_rate": 9.920896445924373e-05,
+      "loss": 0.2917,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4043416976928711,
+      "learning_rate": 9.919530398702917e-05,
+      "loss": 0.3639,
+      "step": 181
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4486939311027527,
+      "learning_rate": 9.918152752708438e-05,
+      "loss": 0.3563,
+      "step": 182
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.428647518157959,
+      "learning_rate": 9.916763511189009e-05,
+      "loss": 0.2876,
+      "step": 183
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.387262761592865,
+      "learning_rate": 9.915362677420044e-05,
+      "loss": 0.2808,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5288756489753723,
+      "learning_rate": 9.913950254704291e-05,
+      "loss": 0.37,
+      "step": 185
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.45147794485092163,
+      "learning_rate": 9.912526246371814e-05,
+      "loss": 0.334,
+      "step": 186
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.392228901386261,
+      "learning_rate": 9.911090655779997e-05,
+      "loss": 0.3256,
+      "step": 187
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.43259549140930176,
+      "learning_rate": 9.909643486313533e-05,
+      "loss": 0.2784,
+      "step": 188
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.49285900592803955,
+      "learning_rate": 9.908184741384412e-05,
+      "loss": 0.3567,
+      "step": 189
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.42180612683296204,
+      "learning_rate": 9.906714424431913e-05,
+      "loss": 0.2176,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4453238248825073,
+      "learning_rate": 9.905232538922605e-05,
+      "loss": 0.2959,
+      "step": 191
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.43805554509162903,
+      "learning_rate": 9.903739088350325e-05,
+      "loss": 0.3074,
+      "step": 192
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3552888333797455,
+      "learning_rate": 9.902234076236181e-05,
+      "loss": 0.1883,
+      "step": 193
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.43793925642967224,
+      "learning_rate": 9.900717506128541e-05,
+      "loss": 0.278,
+      "step": 194
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.49150022864341736,
+      "learning_rate": 9.899189381603018e-05,
+      "loss": 0.3677,
+      "step": 195
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5191401839256287,
+      "learning_rate": 9.897649706262473e-05,
+      "loss": 0.2823,
+      "step": 196
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.49344751238822937,
+      "learning_rate": 9.896098483736995e-05,
+      "loss": 0.3471,
+      "step": 197
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.47374844551086426,
+      "learning_rate": 9.894535717683902e-05,
+      "loss": 0.3427,
+      "step": 198
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3957318663597107,
+      "learning_rate": 9.892961411787724e-05,
+      "loss": 0.2361,
+      "step": 199
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4086719751358032,
+      "learning_rate": 9.891375569760204e-05,
+      "loss": 0.2792,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.3120017349720001,
+      "eval_runtime": 198.2498,
+      "eval_samples_per_second": 1.72,
+      "eval_steps_per_second": 0.217,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4237310290336609,
+      "learning_rate": 9.889778195340278e-05,
+      "loss": 0.295,
+      "step": 201
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.47402215003967285,
+      "learning_rate": 9.888169292294076e-05,
+      "loss": 0.3112,
+      "step": 202
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.48910877108573914,
+      "learning_rate": 9.886548864414907e-05,
+      "loss": 0.3444,
+      "step": 203
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.41279616951942444,
+      "learning_rate": 9.884916915523251e-05,
+      "loss": 0.2892,
+      "step": 204
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4935925304889679,
+      "learning_rate": 9.883273449466754e-05,
+      "loss": 0.3918,
+      "step": 205
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4988994002342224,
+      "learning_rate": 9.881618470120215e-05,
+      "loss": 0.3443,
+      "step": 206
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7123827338218689,
+      "learning_rate": 9.879951981385578e-05,
+      "loss": 0.3557,
+      "step": 207
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.37120532989501953,
+      "learning_rate": 9.87827398719192e-05,
+      "loss": 0.1954,
+      "step": 208
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.435150682926178,
+      "learning_rate": 9.876584491495447e-05,
+      "loss": 0.3396,
+      "step": 209
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3208915591239929,
+      "learning_rate": 9.874883498279484e-05,
+      "loss": 0.2126,
+      "step": 210
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.45818355679512024,
+      "learning_rate": 9.87317101155446e-05,
+      "loss": 0.3579,
+      "step": 211
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.33168700337409973,
+      "learning_rate": 9.871447035357903e-05,
+      "loss": 0.2481,
+      "step": 212
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3689310550689697,
+      "learning_rate": 9.869711573754433e-05,
+      "loss": 0.2605,
+      "step": 213
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3976489007472992,
+      "learning_rate": 9.867964630835743e-05,
+      "loss": 0.318,
+      "step": 214
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3683297336101532,
+      "learning_rate": 9.866206210720599e-05,
+      "loss": 0.2888,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5090157985687256,
+      "learning_rate": 9.864436317554831e-05,
+      "loss": 0.3836,
+      "step": 216
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.449663907289505,
+      "learning_rate": 9.862654955511309e-05,
+      "loss": 0.3996,
+      "step": 217
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.48502326011657715,
+      "learning_rate": 9.860862128789953e-05,
+      "loss": 0.411,
+      "step": 218
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4194256067276001,
+      "learning_rate": 9.859057841617709e-05,
+      "loss": 0.3306,
+      "step": 219
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3323703408241272,
+      "learning_rate": 9.857242098248542e-05,
+      "loss": 0.2191,
+      "step": 220
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.37762174010276794,
+      "learning_rate": 9.855414902963429e-05,
+      "loss": 0.2605,
+      "step": 221
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.44715696573257446,
+      "learning_rate": 9.853576260070347e-05,
+      "loss": 0.3587,
+      "step": 222
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.42095401883125305,
+      "learning_rate": 9.851726173904264e-05,
+      "loss": 0.248,
+      "step": 223
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4548057019710541,
+      "learning_rate": 9.849864648827125e-05,
+      "loss": 0.3896,
+      "step": 224
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4253760874271393,
+      "learning_rate": 9.847991689227847e-05,
+      "loss": 0.3287,
+      "step": 225
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4457933008670807,
+      "learning_rate": 9.846107299522304e-05,
+      "loss": 0.2508,
+      "step": 226
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.33387988805770874,
+      "learning_rate": 9.844211484153325e-05,
+      "loss": 0.2338,
+      "step": 227
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4621600806713104,
+      "learning_rate": 9.842304247590667e-05,
+      "loss": 0.3367,
+      "step": 228
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5213073492050171,
+      "learning_rate": 9.840385594331022e-05,
+      "loss": 0.3071,
+      "step": 229
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4290790557861328,
+      "learning_rate": 9.838455528897998e-05,
+      "loss": 0.295,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4603792428970337,
+      "learning_rate": 9.836514055842107e-05,
+      "loss": 0.3561,
+      "step": 231
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3734110891819,
+      "learning_rate": 9.834561179740762e-05,
+      "loss": 0.2762,
+      "step": 232
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4285050928592682,
+      "learning_rate": 9.832596905198254e-05,
+      "loss": 0.3149,
+      "step": 233
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.45419713854789734,
+      "learning_rate": 9.830621236845754e-05,
+      "loss": 0.2854,
+      "step": 234
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42688748240470886,
+      "learning_rate": 9.828634179341292e-05,
+      "loss": 0.2779,
+      "step": 235
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5902850031852722,
+      "learning_rate": 9.826635737369751e-05,
+      "loss": 0.4377,
+      "step": 236
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4068024754524231,
+      "learning_rate": 9.82462591564286e-05,
+      "loss": 0.2791,
+      "step": 237
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4278412461280823,
+      "learning_rate": 9.82260471889917e-05,
+      "loss": 0.2996,
+      "step": 238
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4042939245700836,
+      "learning_rate": 9.82057215190406e-05,
+      "loss": 0.2985,
+      "step": 239
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.47565755248069763,
+      "learning_rate": 9.818528219449705e-05,
+      "loss": 0.3374,
+      "step": 240
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.46745574474334717,
+      "learning_rate": 9.816472926355086e-05,
+      "loss": 0.3852,
+      "step": 241
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5099470019340515,
+      "learning_rate": 9.814406277465968e-05,
+      "loss": 0.4441,
+      "step": 242
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41835546493530273,
+      "learning_rate": 9.812328277654888e-05,
+      "loss": 0.2994,
+      "step": 243
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4324185848236084,
+      "learning_rate": 9.810238931821138e-05,
+      "loss": 0.2731,
+      "step": 244
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.48156309127807617,
+      "learning_rate": 9.808138244890775e-05,
+      "loss": 0.3497,
+      "step": 245
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.38469818234443665,
+      "learning_rate": 9.806026221816581e-05,
+      "loss": 0.3195,
+      "step": 246
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4838896095752716,
+      "learning_rate": 9.803902867578075e-05,
+      "loss": 0.4426,
+      "step": 247
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4115101993083954,
+      "learning_rate": 9.801768187181486e-05,
+      "loss": 0.315,
+      "step": 248
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3617348372936249,
+      "learning_rate": 9.799622185659747e-05,
+      "loss": 0.2437,
+      "step": 249
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.37255430221557617,
+      "learning_rate": 9.797464868072488e-05,
+      "loss": 0.271,
+      "step": 250
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4224826693534851,
+      "learning_rate": 9.795296239506012e-05,
+      "loss": 0.3425,
+      "step": 251
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4121735990047455,
+      "learning_rate": 9.793116305073292e-05,
+      "loss": 0.3329,
+      "step": 252
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3529704809188843,
+      "learning_rate": 9.79092506991396e-05,
+      "loss": 0.2316,
+      "step": 253
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.43581584095954895,
+      "learning_rate": 9.788722539194291e-05,
+      "loss": 0.3546,
+      "step": 254
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3596087098121643,
+      "learning_rate": 9.786508718107185e-05,
+      "loss": 0.2266,
+      "step": 255
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.42186716198921204,
+      "learning_rate": 9.784283611872169e-05,
+      "loss": 0.2847,
+      "step": 256
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4113231301307678,
+      "learning_rate": 9.782047225735376e-05,
+      "loss": 0.2511,
+      "step": 257
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.40256384015083313,
+      "learning_rate": 9.77979956496953e-05,
+      "loss": 0.2598,
+      "step": 258
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4532873034477234,
+      "learning_rate": 9.777540634873938e-05,
+      "loss": 0.3514,
+      "step": 259
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5714769959449768,
+      "learning_rate": 9.77527044077448e-05,
+      "loss": 0.4927,
+      "step": 260
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4540325701236725,
+      "learning_rate": 9.772988988023588e-05,
+      "loss": 0.375,
+      "step": 261
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.391695111989975,
+      "learning_rate": 9.770696282000244e-05,
+      "loss": 0.2519,
+      "step": 262
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4740663766860962,
+      "learning_rate": 9.768392328109958e-05,
+      "loss": 0.3972,
+      "step": 263
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5314821004867554,
+      "learning_rate": 9.766077131784763e-05,
+      "loss": 0.3422,
+      "step": 264
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.40060651302337646,
+      "learning_rate": 9.763750698483191e-05,
+      "loss": 0.325,
+      "step": 265
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4043687880039215,
+      "learning_rate": 9.761413033690276e-05,
+      "loss": 0.1728,
+      "step": 266
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5536463856697083,
+      "learning_rate": 9.759064142917526e-05,
+      "loss": 0.3692,
+      "step": 267
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.49006491899490356,
+      "learning_rate": 9.756704031702918e-05,
+      "loss": 0.3886,
+      "step": 268
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.42059868574142456,
+      "learning_rate": 9.754332705610889e-05,
+      "loss": 0.2734,
+      "step": 269
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4724079966545105,
+      "learning_rate": 9.751950170232309e-05,
+      "loss": 0.3028,
+      "step": 270
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.41685357689857483,
+      "learning_rate": 9.74955643118448e-05,
+      "loss": 0.2857,
+      "step": 271
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4813288152217865,
+      "learning_rate": 9.74715149411112e-05,
+      "loss": 0.302,
+      "step": 272
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.45547735691070557,
+      "learning_rate": 9.744735364682346e-05,
+      "loss": 0.3333,
+      "step": 273
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6381693482398987,
+      "learning_rate": 9.742308048594665e-05,
+      "loss": 0.3558,
+      "step": 274
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.49117156863212585,
+      "learning_rate": 9.739869551570959e-05,
+      "loss": 0.3754,
+      "step": 275
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4352806508541107,
+      "learning_rate": 9.73741987936047e-05,
+      "loss": 0.257,
+      "step": 276
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4876728057861328,
+      "learning_rate": 9.734959037738787e-05,
+      "loss": 0.3257,
+      "step": 277
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5180380344390869,
+      "learning_rate": 9.732487032507837e-05,
+      "loss": 0.3041,
+      "step": 278
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4687010943889618,
+      "learning_rate": 9.730003869495863e-05,
+      "loss": 0.3284,
+      "step": 279
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.474204957485199,
+      "learning_rate": 9.727509554557417e-05,
+      "loss": 0.2296,
+      "step": 280
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4333823621273041,
+      "learning_rate": 9.725004093573342e-05,
+      "loss": 0.3164,
+      "step": 281
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.47495847940444946,
+      "learning_rate": 9.722487492450764e-05,
+      "loss": 0.3497,
+      "step": 282
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4860828220844269,
+      "learning_rate": 9.719959757123071e-05,
+      "loss": 0.3787,
+      "step": 283
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4903152883052826,
+      "learning_rate": 9.717420893549902e-05,
+      "loss": 0.4795,
+      "step": 284
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4164125323295593,
+      "learning_rate": 9.714870907717133e-05,
+      "loss": 0.3688,
+      "step": 285
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.46644559502601624,
+      "learning_rate": 9.712309805636863e-05,
+      "loss": 0.407,
+      "step": 286
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4093582034111023,
+      "learning_rate": 9.709737593347403e-05,
+      "loss": 0.3007,
+      "step": 287
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5839386582374573,
+      "learning_rate": 9.707154276913256e-05,
+      "loss": 0.4525,
+      "step": 288
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5002758502960205,
+      "learning_rate": 9.704559862425102e-05,
+      "loss": 0.3956,
+      "step": 289
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4330982565879822,
+      "learning_rate": 9.701954355999792e-05,
+      "loss": 0.2816,
+      "step": 290
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.47351762652397156,
+      "learning_rate": 9.699337763780326e-05,
+      "loss": 0.3837,
+      "step": 291
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.48483800888061523,
+      "learning_rate": 9.696710091935841e-05,
+      "loss": 0.4345,
+      "step": 292
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.43565940856933594,
+      "learning_rate": 9.694071346661599e-05,
+      "loss": 0.345,
+      "step": 293
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4082830548286438,
+      "learning_rate": 9.691421534178966e-05,
+      "loss": 0.2585,
+      "step": 294
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5202715396881104,
+      "learning_rate": 9.688760660735402e-05,
+      "loss": 0.3441,
+      "step": 295
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4059908390045166,
+      "learning_rate": 9.686088732604449e-05,
+      "loss": 0.2605,
+      "step": 296
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3188331425189972,
+      "learning_rate": 9.683405756085708e-05,
+      "loss": 0.1557,
+      "step": 297
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4404698610305786,
+      "learning_rate": 9.68071173750483e-05,
+      "loss": 0.3832,
+      "step": 298
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.41592538356781006,
+      "learning_rate": 9.678006683213503e-05,
+      "loss": 0.326,
+      "step": 299
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.41149288415908813,
+      "learning_rate": 9.675290599589429e-05,
+      "loss": 0.2943,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.3020077049732208,
+      "eval_runtime": 196.7612,
+      "eval_samples_per_second": 1.733,
+      "eval_steps_per_second": 0.219,
+      "step": 300
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.48964405059814453,
+      "learning_rate": 9.672563493036318e-05,
+      "loss": 0.3514,
+      "step": 301
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.47688812017440796,
+      "learning_rate": 9.669825369983865e-05,
+      "loss": 0.3215,
+      "step": 302
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.38745707273483276,
+      "learning_rate": 9.667076236887743e-05,
+      "loss": 0.2291,
+      "step": 303
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5546700954437256,
+      "learning_rate": 9.664316100229578e-05,
+      "loss": 0.3636,
+      "step": 304
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.47053879499435425,
+      "learning_rate": 9.661544966516945e-05,
+      "loss": 0.3544,
+      "step": 305
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.48200368881225586,
+      "learning_rate": 9.658762842283343e-05,
+      "loss": 0.3435,
+      "step": 306
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4663654565811157,
+      "learning_rate": 9.655969734088184e-05,
+      "loss": 0.3836,
+      "step": 307
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5240476727485657,
+      "learning_rate": 9.653165648516776e-05,
+      "loss": 0.409,
+      "step": 308
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.47078466415405273,
+      "learning_rate": 9.65035059218031e-05,
+      "loss": 0.4209,
+      "step": 309
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.38707101345062256,
+      "learning_rate": 9.647524571715843e-05,
+      "loss": 0.2406,
+      "step": 310
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4670199155807495,
+      "learning_rate": 9.644687593786282e-05,
+      "loss": 0.3189,
+      "step": 311
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4175633192062378,
+      "learning_rate": 9.641839665080363e-05,
+      "loss": 0.2792,
+      "step": 312
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40232059359550476,
+      "learning_rate": 9.638980792312651e-05,
+      "loss": 0.2571,
+      "step": 313
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.40628305077552795,
+      "learning_rate": 9.636110982223505e-05,
+      "loss": 0.2905,
+      "step": 314
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.44698962569236755,
+      "learning_rate": 9.633230241579076e-05,
+      "loss": 0.3521,
+      "step": 315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4806763529777527,
+      "learning_rate": 9.630338577171282e-05,
+      "loss": 0.3101,
+      "step": 316
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4798329174518585,
+      "learning_rate": 9.627435995817799e-05,
+      "loss": 0.3324,
+      "step": 317
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4957510232925415,
+      "learning_rate": 9.624522504362039e-05,
+      "loss": 0.3177,
+      "step": 318
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.38843831419944763,
+      "learning_rate": 9.621598109673142e-05,
+      "loss": 0.2523,
+      "step": 319
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5394728779792786,
+      "learning_rate": 9.618662818645948e-05,
+      "loss": 0.4743,
+      "step": 320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5877221822738647,
+      "learning_rate": 9.615716638200993e-05,
+      "loss": 0.4485,
+      "step": 321
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.42556509375572205,
+      "learning_rate": 9.612759575284483e-05,
+      "loss": 0.3206,
+      "step": 322
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3942933976650238,
+      "learning_rate": 9.609791636868281e-05,
+      "loss": 0.2973,
+      "step": 323
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.38953080773353577,
+      "learning_rate": 9.606812829949896e-05,
+      "loss": 0.2593,
+      "step": 324
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.35556721687316895,
+      "learning_rate": 9.603823161552457e-05,
+      "loss": 0.2268,
+      "step": 325
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.41105249524116516,
+      "learning_rate": 9.600822638724705e-05,
+      "loss": 0.2727,
+      "step": 326
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4370076060295105,
+      "learning_rate": 9.597811268540968e-05,
+      "loss": 0.2921,
+      "step": 327
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4090844988822937,
+      "learning_rate": 9.594789058101153e-05,
+      "loss": 0.3453,
+      "step": 328
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5412084460258484,
+      "learning_rate": 9.591756014530722e-05,
+      "loss": 0.3819,
+      "step": 329
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4325799345970154,
+      "learning_rate": 9.588712144980681e-05,
+      "loss": 0.2918,
+      "step": 330
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.412055104970932,
+      "learning_rate": 9.585657456627556e-05,
+      "loss": 0.3288,
+      "step": 331
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.43470674753189087,
+      "learning_rate": 9.582591956673387e-05,
+      "loss": 0.3594,
+      "step": 332
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.42225292325019836,
+      "learning_rate": 9.579515652345698e-05,
+      "loss": 0.3232,
+      "step": 333
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4531938433647156,
+      "learning_rate": 9.576428550897489e-05,
+      "loss": 0.3352,
+      "step": 334
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.43436199426651,
+      "learning_rate": 9.57333065960722e-05,
+      "loss": 0.3303,
+      "step": 335
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.47585529088974,
+      "learning_rate": 9.570221985778785e-05,
+      "loss": 0.3899,
+      "step": 336
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4278468191623688,
+      "learning_rate": 9.567102536741501e-05,
+      "loss": 0.2915,
+      "step": 337
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5330191850662231,
+      "learning_rate": 9.563972319850092e-05,
+      "loss": 0.3293,
+      "step": 338
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.49070894718170166,
+      "learning_rate": 9.560831342484667e-05,
+      "loss": 0.3825,
+      "step": 339
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4945015609264374,
+      "learning_rate": 9.557679612050708e-05,
+      "loss": 0.3208,
+      "step": 340
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3731319010257721,
+      "learning_rate": 9.554517135979043e-05,
+      "loss": 0.2647,
+      "step": 341
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4276984930038452,
+      "learning_rate": 9.551343921725843e-05,
+      "loss": 0.3548,
+      "step": 342
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4000339210033417,
+      "learning_rate": 9.548159976772592e-05,
+      "loss": 0.2391,
+      "step": 343
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.40203213691711426,
+      "learning_rate": 9.544965308626074e-05,
+      "loss": 0.2585,
+      "step": 344
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.49098315834999084,
+      "learning_rate": 9.541759924818357e-05,
+      "loss": 0.3881,
+      "step": 345
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5103345513343811,
+      "learning_rate": 9.538543832906773e-05,
+      "loss": 0.4239,
+      "step": 346
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.372951477766037,
+      "learning_rate": 9.535317040473895e-05,
+      "loss": 0.253,
+      "step": 347
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.33518075942993164,
+      "learning_rate": 9.53207955512753e-05,
+      "loss": 0.1784,
+      "step": 348
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4415912628173828,
+      "learning_rate": 9.528831384500698e-05,
+      "loss": 0.3254,
+      "step": 349
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.44483518600463867,
+      "learning_rate": 9.525572536251607e-05,
+      "loss": 0.287,
+      "step": 350
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4194684326648712,
+      "learning_rate": 9.52230301806364e-05,
+      "loss": 0.314,
+      "step": 351
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.44148579239845276,
+      "learning_rate": 9.519022837645337e-05,
+      "loss": 0.3175,
+      "step": 352
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4267377555370331,
+      "learning_rate": 9.51573200273038e-05,
+      "loss": 0.2793,
+      "step": 353
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.45702677965164185,
+      "learning_rate": 9.512430521077565e-05,
+      "loss": 0.3318,
+      "step": 354
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3175434470176697,
+      "learning_rate": 9.509118400470793e-05,
+      "loss": 0.188,
+      "step": 355
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40171679854393005,
+      "learning_rate": 9.505795648719048e-05,
+      "loss": 0.2964,
+      "step": 356
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3963138461112976,
+      "learning_rate": 9.50246227365638e-05,
+      "loss": 0.26,
+      "step": 357
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.475111722946167,
+      "learning_rate": 9.499118283141887e-05,
+      "loss": 0.3673,
+      "step": 358
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4654029309749603,
+      "learning_rate": 9.495763685059688e-05,
+      "loss": 0.3721,
+      "step": 359
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.45381075143814087,
+      "learning_rate": 9.492398487318921e-05,
+      "loss": 0.3059,
+      "step": 360
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5407252907752991,
+      "learning_rate": 9.489022697853709e-05,
+      "loss": 0.3051,
+      "step": 361
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4436393082141876,
+      "learning_rate": 9.485636324623146e-05,
+      "loss": 0.3265,
+      "step": 362
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.48739132285118103,
+      "learning_rate": 9.482239375611282e-05,
+      "loss": 0.3045,
+      "step": 363
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5032399892807007,
+      "learning_rate": 9.478831858827104e-05,
+      "loss": 0.3267,
+      "step": 364
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3174649775028229,
+      "learning_rate": 9.475413782304507e-05,
+      "loss": 0.1427,
+      "step": 365
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.430126816034317,
+      "learning_rate": 9.471985154102292e-05,
+      "loss": 0.264,
+      "step": 366
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.46137863397598267,
+      "learning_rate": 9.468545982304132e-05,
+      "loss": 0.2683,
+      "step": 367
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.44973912835121155,
+      "learning_rate": 9.465096275018556e-05,
+      "loss": 0.2504,
+      "step": 368
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3683367967605591,
+      "learning_rate": 9.46163604037894e-05,
+      "loss": 0.2321,
+      "step": 369
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4687761664390564,
+      "learning_rate": 9.458165286543476e-05,
+      "loss": 0.3952,
+      "step": 370
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4155949056148529,
+      "learning_rate": 9.454684021695157e-05,
+      "loss": 0.312,
+      "step": 371
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4418841004371643,
+      "learning_rate": 9.451192254041758e-05,
+      "loss": 0.3289,
+      "step": 372
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3964237868785858,
+      "learning_rate": 9.447689991815818e-05,
+      "loss": 0.2284,
+      "step": 373
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.41220808029174805,
+      "learning_rate": 9.444177243274618e-05,
+      "loss": 0.294,
+      "step": 374
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.43764981627464294,
+      "learning_rate": 9.440654016700161e-05,
+      "loss": 0.3007,
+      "step": 375
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3570357859134674,
+      "learning_rate": 9.437120320399158e-05,
+      "loss": 0.2227,
+      "step": 376
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4318711757659912,
+      "learning_rate": 9.433576162703e-05,
+      "loss": 0.3641,
+      "step": 377
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.459570050239563,
+      "learning_rate": 9.430021551967744e-05,
+      "loss": 0.3101,
+      "step": 378
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4114058017730713,
+      "learning_rate": 9.426456496574096e-05,
+      "loss": 0.2665,
+      "step": 379
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.36416831612586975,
+      "learning_rate": 9.422881004927381e-05,
+      "loss": 0.2135,
+      "step": 380
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4428749978542328,
+      "learning_rate": 9.419295085457536e-05,
+      "loss": 0.3626,
+      "step": 381
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.26782330870628357,
+      "learning_rate": 9.415698746619079e-05,
+      "loss": 0.1652,
+      "step": 382
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.35025525093078613,
+      "learning_rate": 9.412091996891096e-05,
+      "loss": 0.2179,
+      "step": 383
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.49769869446754456,
+      "learning_rate": 9.408474844777217e-05,
+      "loss": 0.4192,
+      "step": 384
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.44601088762283325,
+      "learning_rate": 9.404847298805601e-05,
+      "loss": 0.3209,
+      "step": 385
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4931808412075043,
+      "learning_rate": 9.401209367528908e-05,
+      "loss": 0.4071,
+      "step": 386
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.38014572858810425,
+      "learning_rate": 9.397561059524285e-05,
+      "loss": 0.2517,
+      "step": 387
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40229374170303345,
+      "learning_rate": 9.393902383393347e-05,
+      "loss": 0.3056,
+      "step": 388
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4664427936077118,
+      "learning_rate": 9.39023334776215e-05,
+      "loss": 0.2903,
+      "step": 389
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.45850861072540283,
+      "learning_rate": 9.386553961281179e-05,
+      "loss": 0.3753,
+      "step": 390
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3542648255825043,
+      "learning_rate": 9.382864232625322e-05,
+      "loss": 0.2094,
+      "step": 391
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4940478205680847,
+      "learning_rate": 9.379164170493844e-05,
+      "loss": 0.3666,
+      "step": 392
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.37893354892730713,
+      "learning_rate": 9.375453783610382e-05,
+      "loss": 0.2583,
+      "step": 393
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3733774721622467,
+      "learning_rate": 9.371733080722911e-05,
+      "loss": 0.279,
+      "step": 394
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3752308487892151,
+      "learning_rate": 9.36800207060373e-05,
+      "loss": 0.3112,
+      "step": 395
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.383168488740921,
+      "learning_rate": 9.36426076204944e-05,
+      "loss": 0.2145,
+      "step": 396
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3427211344242096,
+      "learning_rate": 9.360509163880921e-05,
+      "loss": 0.1918,
+      "step": 397
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.39450734853744507,
+      "learning_rate": 9.35674728494331e-05,
+      "loss": 0.2871,
+      "step": 398
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4286922216415405,
+      "learning_rate": 9.35297513410599e-05,
+      "loss": 0.3246,
+      "step": 399
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3923003375530243,
+      "learning_rate": 9.349192720262555e-05,
+      "loss": 0.2246,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.29848504066467285,
+      "eval_runtime": 195.2879,
+      "eval_samples_per_second": 1.746,
+      "eval_steps_per_second": 0.22,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5048094987869263,
+      "learning_rate": 9.345400052330802e-05,
+      "loss": 0.3582,
+      "step": 401
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.45082950592041016,
+      "learning_rate": 9.341597139252698e-05,
+      "loss": 0.3352,
+      "step": 402
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.42727285623550415,
+      "learning_rate": 9.337783989994371e-05,
+      "loss": 0.336,
+      "step": 403
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.41211187839508057,
+      "learning_rate": 9.333960613546079e-05,
+      "loss": 0.2273,
+      "step": 404
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3849753737449646,
+      "learning_rate": 9.330127018922194e-05,
+      "loss": 0.2385,
+      "step": 405
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.42667919397354126,
+      "learning_rate": 9.326283215161178e-05,
+      "loss": 0.3392,
+      "step": 406
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.47514185309410095,
+      "learning_rate": 9.322429211325566e-05,
+      "loss": 0.3759,
+      "step": 407
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4542549252510071,
+      "learning_rate": 9.31856501650194e-05,
+      "loss": 0.3713,
+      "step": 408
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.42247703671455383,
+      "learning_rate": 9.314690639800906e-05,
+      "loss": 0.3052,
+      "step": 409
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.43194207549095154,
+      "learning_rate": 9.310806090357082e-05,
+      "loss": 0.35,
+      "step": 410
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5062002539634705,
+      "learning_rate": 9.306911377329066e-05,
+      "loss": 0.3658,
+      "step": 411
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.37047699093818665,
+      "learning_rate": 9.303006509899418e-05,
+      "loss": 0.2302,
+      "step": 412
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.37179237604141235,
+      "learning_rate": 9.299091497274647e-05,
+      "loss": 0.3088,
+      "step": 413
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5352928638458252,
+      "learning_rate": 9.29516634868517e-05,
+      "loss": 0.3772,
+      "step": 414
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4247976839542389,
+      "learning_rate": 9.291231073385307e-05,
+      "loss": 0.2284,
+      "step": 415
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4145027697086334,
+      "learning_rate": 9.287285680653254e-05,
+      "loss": 0.269,
+      "step": 416
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4691685736179352,
+      "learning_rate": 9.283330179791063e-05,
+      "loss": 0.331,
+      "step": 417
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5112786293029785,
+      "learning_rate": 9.279364580124614e-05,
+      "loss": 0.3504,
+      "step": 418
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4482985734939575,
+      "learning_rate": 9.275388891003596e-05,
+      "loss": 0.2665,
+      "step": 419
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.44037336111068726,
+      "learning_rate": 9.27140312180149e-05,
+      "loss": 0.3481,
+      "step": 420
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4734729528427124,
+      "learning_rate": 9.267407281915541e-05,
+      "loss": 0.3263,
+      "step": 421
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.47684046626091003,
+      "learning_rate": 9.263401380766739e-05,
+      "loss": 0.3454,
+      "step": 422
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4239022433757782,
+      "learning_rate": 9.25938542779979e-05,
+      "loss": 0.2782,
+      "step": 423
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.46097472310066223,
+      "learning_rate": 9.255359432483105e-05,
+      "loss": 0.3407,
+      "step": 424
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4574018120765686,
+      "learning_rate": 9.251323404308774e-05,
+      "loss": 0.2955,
+      "step": 425
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5474076867103577,
+      "learning_rate": 9.247277352792533e-05,
+      "loss": 0.4942,
+      "step": 426
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4703276753425598,
+      "learning_rate": 9.243221287473756e-05,
+      "loss": 0.388,
+      "step": 427
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4886913597583771,
+      "learning_rate": 9.239155217915422e-05,
+      "loss": 0.3884,
+      "step": 428
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.45662686228752136,
+      "learning_rate": 9.235079153704107e-05,
+      "loss": 0.3941,
+      "step": 429
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4769531190395355,
+      "learning_rate": 9.230993104449939e-05,
+      "loss": 0.3388,
+      "step": 430
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.42851901054382324,
+      "learning_rate": 9.226897079786593e-05,
+      "loss": 0.2974,
+      "step": 431
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4670259952545166,
+      "learning_rate": 9.222791089371265e-05,
+      "loss": 0.3391,
+      "step": 432
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4410526752471924,
+      "learning_rate": 9.218675142884647e-05,
+      "loss": 0.302,
+      "step": 433
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3526303172111511,
+      "learning_rate": 9.214549250030899e-05,
+      "loss": 0.2067,
+      "step": 434
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4363113343715668,
+      "learning_rate": 9.210413420537638e-05,
+      "loss": 0.3364,
+      "step": 435
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4459896981716156,
+      "learning_rate": 9.206267664155907e-05,
+      "loss": 0.3277,
+      "step": 436
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5150530934333801,
+      "learning_rate": 9.20211199066015e-05,
+      "loss": 0.3753,
+      "step": 437
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5036837458610535,
+      "learning_rate": 9.197946409848194e-05,
+      "loss": 0.3015,
+      "step": 438
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6625205278396606,
+      "learning_rate": 9.19377093154123e-05,
+      "loss": 0.184,
+      "step": 439
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.45180633664131165,
+      "learning_rate": 9.189585565583778e-05,
+      "loss": 0.3118,
+      "step": 440
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.44038087129592896,
+      "learning_rate": 9.185390321843672e-05,
+      "loss": 0.3085,
+      "step": 441
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.46567392349243164,
+      "learning_rate": 9.181185210212034e-05,
+      "loss": 0.2982,
+      "step": 442
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.44771289825439453,
+      "learning_rate": 9.176970240603253e-05,
+      "loss": 0.2934,
+      "step": 443
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.45960354804992676,
+      "learning_rate": 9.172745422954961e-05,
+      "loss": 0.347,
+      "step": 444
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.475410521030426,
+      "learning_rate": 9.168510767228007e-05,
+      "loss": 0.3743,
+      "step": 445
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5116153359413147,
+      "learning_rate": 9.164266283406433e-05,
+      "loss": 0.4004,
+      "step": 446
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.395393043756485,
+      "learning_rate": 9.160011981497457e-05,
+      "loss": 0.2926,
+      "step": 447
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.40677595138549805,
+      "learning_rate": 9.155747871531444e-05,
+      "loss": 0.2975,
+      "step": 448
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.34814658761024475,
+      "learning_rate": 9.151473963561883e-05,
+      "loss": 0.2043,
+      "step": 449
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.51041579246521,
+      "learning_rate": 9.14719026766536e-05,
+      "loss": 0.3483,
+      "step": 450
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4424203932285309,
+      "learning_rate": 9.142896793941546e-05,
+      "loss": 0.3222,
+      "step": 451
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5328881740570068,
+      "learning_rate": 9.13859355251316e-05,
+      "loss": 0.4283,
+      "step": 452
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4070656895637512,
+      "learning_rate": 9.134280553525946e-05,
+      "loss": 0.2327,
+      "step": 453
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5129764080047607,
+      "learning_rate": 9.129957807148666e-05,
+      "loss": 0.358,
+      "step": 454
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5004768967628479,
+      "learning_rate": 9.12562532357305e-05,
+      "loss": 0.3207,
+      "step": 455
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4688027501106262,
+      "learning_rate": 9.121283113013795e-05,
+      "loss": 0.3124,
+      "step": 456
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.500098466873169,
+      "learning_rate": 9.116931185708523e-05,
+      "loss": 0.3502,
+      "step": 457
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4750964343547821,
+      "learning_rate": 9.112569551917774e-05,
+      "loss": 0.3692,
+      "step": 458
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.47686442732810974,
+      "learning_rate": 9.108198221924966e-05,
+      "loss": 0.3375,
+      "step": 459
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.35196036100387573,
+      "learning_rate": 9.103817206036382e-05,
+      "loss": 0.2595,
+      "step": 460
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4437873065471649,
+      "learning_rate": 9.09942651458114e-05,
+      "loss": 0.3587,
+      "step": 461
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5091555714607239,
+      "learning_rate": 9.095026157911166e-05,
+      "loss": 0.3663,
+      "step": 462
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4090098738670349,
+      "learning_rate": 9.090616146401183e-05,
+      "loss": 0.3196,
+      "step": 463
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.363640159368515,
+      "learning_rate": 9.086196490448668e-05,
+      "loss": 0.2584,
+      "step": 464
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.48382431268692017,
+      "learning_rate": 9.081767200473842e-05,
+      "loss": 0.3539,
+      "step": 465
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4430081844329834,
+      "learning_rate": 9.077328286919638e-05,
+      "loss": 0.3058,
+      "step": 466
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3796209990978241,
+      "learning_rate": 9.072879760251679e-05,
+      "loss": 0.2381,
+      "step": 467
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4326118528842926,
+      "learning_rate": 9.068421630958254e-05,
+      "loss": 0.3087,
+      "step": 468
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.40796566009521484,
+      "learning_rate": 9.06395390955029e-05,
+      "loss": 0.2317,
+      "step": 469
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.47974613308906555,
+      "learning_rate": 9.059476606561328e-05,
+      "loss": 0.3867,
+      "step": 470
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4144863188266754,
+      "learning_rate": 9.054989732547506e-05,
+      "loss": 0.3107,
+      "step": 471
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.43266648054122925,
+      "learning_rate": 9.050493298087523e-05,
+      "loss": 0.2965,
+      "step": 472
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.44545048475265503,
+      "learning_rate": 9.045987313782616e-05,
+      "loss": 0.3455,
+      "step": 473
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.43648597598075867,
+      "learning_rate": 9.041471790256542e-05,
+      "loss": 0.3373,
+      "step": 474
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3785746097564697,
+      "learning_rate": 9.036946738155548e-05,
+      "loss": 0.2345,
+      "step": 475
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.43230387568473816,
+      "learning_rate": 9.032412168148345e-05,
+      "loss": 0.2499,
+      "step": 476
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.48966890573501587,
+      "learning_rate": 9.027868090926088e-05,
+      "loss": 0.3727,
+      "step": 477
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.42558932304382324,
+      "learning_rate": 9.02331451720234e-05,
+      "loss": 0.3276,
+      "step": 478
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4567824900150299,
+      "learning_rate": 9.018751457713062e-05,
+      "loss": 0.2971,
+      "step": 479
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3855675160884857,
+      "learning_rate": 9.014178923216572e-05,
+      "loss": 0.2425,
+      "step": 480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4731333553791046,
+      "learning_rate": 9.009596924493535e-05,
+      "loss": 0.3338,
+      "step": 481
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.33461233973503113,
+      "learning_rate": 9.005005472346924e-05,
+      "loss": 0.1347,
+      "step": 482
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4219686686992645,
+      "learning_rate": 9.000404577602002e-05,
+      "loss": 0.3251,
+      "step": 483
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3888438940048218,
+      "learning_rate": 8.995794251106295e-05,
+      "loss": 0.2352,
+      "step": 484
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.46238046884536743,
+      "learning_rate": 8.991174503729567e-05,
+      "loss": 0.2805,
+      "step": 485
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.30291134119033813,
+      "learning_rate": 8.986545346363792e-05,
+      "loss": 0.1787,
+      "step": 486
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3825189173221588,
+      "learning_rate": 8.98190678992313e-05,
+      "loss": 0.2841,
+      "step": 487
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.36861228942871094,
+      "learning_rate": 8.977258845343905e-05,
+      "loss": 0.26,
+      "step": 488
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.39149531722068787,
+      "learning_rate": 8.97260152358457e-05,
+      "loss": 0.2203,
+      "step": 489
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3831264078617096,
+      "learning_rate": 8.967934835625689e-05,
+      "loss": 0.2776,
+      "step": 490
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.43955692648887634,
+      "learning_rate": 8.963258792469908e-05,
+      "loss": 0.3214,
+      "step": 491
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.45454782247543335,
+      "learning_rate": 8.958573405141932e-05,
+      "loss": 0.3706,
+      "step": 492
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.40230172872543335,
+      "learning_rate": 8.953878684688493e-05,
+      "loss": 0.2253,
+      "step": 493
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.47933459281921387,
+      "learning_rate": 8.949174642178332e-05,
+      "loss": 0.3567,
+      "step": 494
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5209866166114807,
+      "learning_rate": 8.944461288702166e-05,
+      "loss": 0.3343,
+      "step": 495
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4351024329662323,
+      "learning_rate": 8.939738635372664e-05,
+      "loss": 0.3008,
+      "step": 496
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3871714174747467,
+      "learning_rate": 8.935006693324423e-05,
+      "loss": 0.2287,
+      "step": 497
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.41198650002479553,
+      "learning_rate": 8.930265473713938e-05,
+      "loss": 0.2851,
+      "step": 498
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5157864093780518,
+      "learning_rate": 8.925514987719579e-05,
+      "loss": 0.3608,
+      "step": 499
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4454146921634674,
+      "learning_rate": 8.920755246541563e-05,
+      "loss": 0.2911,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.29619064927101135,
+      "eval_runtime": 194.3353,
+      "eval_samples_per_second": 1.755,
+      "eval_steps_per_second": 0.221,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.48958247900009155,
+      "learning_rate": 8.915986261401931e-05,
+      "loss": 0.3437,
+      "step": 501
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.425104558467865,
+      "learning_rate": 8.911208043544513e-05,
+      "loss": 0.3323,
+      "step": 502
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.45970556139945984,
+      "learning_rate": 8.906420604234908e-05,
+      "loss": 0.3491,
+      "step": 503
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3786400258541107,
+      "learning_rate": 8.90162395476046e-05,
+      "loss": 0.2639,
+      "step": 504
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.480790913105011,
+      "learning_rate": 8.896818106430224e-05,
+      "loss": 0.3758,
+      "step": 505
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.38297513127326965,
+      "learning_rate": 8.89200307057495e-05,
+      "loss": 0.2547,
+      "step": 506
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.42805609107017517,
+      "learning_rate": 8.887178858547039e-05,
+      "loss": 0.3131,
+      "step": 507
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4482990801334381,
+      "learning_rate": 8.882345481720533e-05,
+      "loss": 0.278,
+      "step": 508
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.42347875237464905,
+      "learning_rate": 8.877502951491082e-05,
+      "loss": 0.3378,
+      "step": 509
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.49425357580184937,
+      "learning_rate": 8.872651279275916e-05,
+      "loss": 0.3434,
+      "step": 510
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.47196516394615173,
+      "learning_rate": 8.867790476513818e-05,
+      "loss": 0.2883,
+      "step": 511
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.505861759185791,
+      "learning_rate": 8.862920554665098e-05,
+      "loss": 0.393,
+      "step": 512
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4252704083919525,
+      "learning_rate": 8.858041525211569e-05,
+      "loss": 0.2965,
+      "step": 513
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.44234761595726013,
+      "learning_rate": 8.853153399656513e-05,
+      "loss": 0.3719,
+      "step": 514
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4896153211593628,
+      "learning_rate": 8.848256189524662e-05,
+      "loss": 0.4154,
+      "step": 515
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4136500656604767,
+      "learning_rate": 8.843349906362163e-05,
+      "loss": 0.2014,
+      "step": 516
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5613704323768616,
+      "learning_rate": 8.838434561736556e-05,
+      "loss": 0.3355,
+      "step": 517
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4642159938812256,
+      "learning_rate": 8.833510167236747e-05,
+      "loss": 0.3206,
+      "step": 518
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.46835535764694214,
+      "learning_rate": 8.828576734472974e-05,
+      "loss": 0.3639,
+      "step": 519
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4356152415275574,
+      "learning_rate": 8.823634275076791e-05,
+      "loss": 0.2432,
+      "step": 520
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4081496298313141,
+      "learning_rate": 8.818682800701029e-05,
+      "loss": 0.2048,
+      "step": 521
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5153374075889587,
+      "learning_rate": 8.813722323019773e-05,
+      "loss": 0.4099,
+      "step": 522
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5132246017456055,
+      "learning_rate": 8.808752853728341e-05,
+      "loss": 0.3144,
+      "step": 523
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.40897759795188904,
+      "learning_rate": 8.803774404543247e-05,
+      "loss": 0.2781,
+      "step": 524
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4927729666233063,
+      "learning_rate": 8.798786987202174e-05,
+      "loss": 0.3219,
+      "step": 525
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3870534598827362,
+      "learning_rate": 8.793790613463955e-05,
+      "loss": 0.2476,
+      "step": 526
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5668165683746338,
+      "learning_rate": 8.788785295108535e-05,
+      "loss": 0.3488,
+      "step": 527
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.47078588604927063,
+      "learning_rate": 8.783771043936948e-05,
+      "loss": 0.3793,
+      "step": 528
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.44360631704330444,
+      "learning_rate": 8.778747871771292e-05,
+      "loss": 0.327,
+      "step": 529
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.46791085600852966,
+      "learning_rate": 8.773715790454695e-05,
+      "loss": 0.2854,
+      "step": 530
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5258610248565674,
+      "learning_rate": 8.768674811851293e-05,
+      "loss": 0.3411,
+      "step": 531
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4382946789264679,
+      "learning_rate": 8.763624947846195e-05,
+      "loss": 0.2667,
+      "step": 532
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.387492299079895,
+      "learning_rate": 8.758566210345462e-05,
+      "loss": 0.2596,
+      "step": 533
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.46629124879837036,
+      "learning_rate": 8.753498611276079e-05,
+      "loss": 0.3385,
+      "step": 534
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5972315073013306,
+      "learning_rate": 8.748422162585914e-05,
+      "loss": 0.446,
+      "step": 535
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3973158001899719,
+      "learning_rate": 8.743336876243712e-05,
+      "loss": 0.2604,
+      "step": 536
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.47990667819976807,
+      "learning_rate": 8.738242764239046e-05,
+      "loss": 0.3856,
+      "step": 537
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4652199447154999,
+      "learning_rate": 8.7331398385823e-05,
+      "loss": 0.2998,
+      "step": 538
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5404556393623352,
+      "learning_rate": 8.728028111304638e-05,
+      "loss": 0.4295,
+      "step": 539
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.44715118408203125,
+      "learning_rate": 8.722907594457974e-05,
+      "loss": 0.3228,
+      "step": 540
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5449092388153076,
+      "learning_rate": 8.717778300114951e-05,
+      "loss": 0.4237,
+      "step": 541
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.33151358366012573,
+      "learning_rate": 8.712640240368899e-05,
+      "loss": 0.2267,
+      "step": 542
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4208608865737915,
+      "learning_rate": 8.707493427333818e-05,
+      "loss": 0.324,
+      "step": 543
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4054446816444397,
+      "learning_rate": 8.702337873144343e-05,
+      "loss": 0.284,
+      "step": 544
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39715248346328735,
+      "learning_rate": 8.697173589955724e-05,
+      "loss": 0.3066,
+      "step": 545
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5018863677978516,
+      "learning_rate": 8.692000589943785e-05,
+      "loss": 0.4032,
+      "step": 546
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.46235954761505127,
+      "learning_rate": 8.686818885304905e-05,
+      "loss": 0.2971,
+      "step": 547
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.43333160877227783,
+      "learning_rate": 8.681628488255986e-05,
+      "loss": 0.3511,
+      "step": 548
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4528295397758484,
+      "learning_rate": 8.676429411034422e-05,
+      "loss": 0.3002,
+      "step": 549
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4215562045574188,
+      "learning_rate": 8.671221665898073e-05,
+      "loss": 0.2893,
+      "step": 550
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39993759989738464,
+      "learning_rate": 8.666005265125236e-05,
+      "loss": 0.2737,
+      "step": 551
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.428223580121994,
+      "learning_rate": 8.660780221014616e-05,
+      "loss": 0.2759,
+      "step": 552
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4598846137523651,
+      "learning_rate": 8.655546545885294e-05,
+      "loss": 0.3232,
+      "step": 553
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3715943694114685,
+      "learning_rate": 8.650304252076703e-05,
+      "loss": 0.2283,
+      "step": 554
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3378869295120239,
+      "learning_rate": 8.645053351948593e-05,
+      "loss": 0.1973,
+      "step": 555
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4649445116519928,
+      "learning_rate": 8.639793857881011e-05,
+      "loss": 0.3267,
+      "step": 556
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.42268118262290955,
+      "learning_rate": 8.634525782274258e-05,
+      "loss": 0.2771,
+      "step": 557
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4799594581127167,
+      "learning_rate": 8.629249137548874e-05,
+      "loss": 0.3882,
+      "step": 558
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3998144865036011,
+      "learning_rate": 8.6239639361456e-05,
+      "loss": 0.2374,
+      "step": 559
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.41236236691474915,
+      "learning_rate": 8.618670190525352e-05,
+      "loss": 0.2856,
+      "step": 560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37495484948158264,
+      "learning_rate": 8.613367913169187e-05,
+      "loss": 0.2334,
+      "step": 561
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4348457157611847,
+      "learning_rate": 8.608057116578283e-05,
+      "loss": 0.3747,
+      "step": 562
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.44270768761634827,
+      "learning_rate": 8.602737813273901e-05,
+      "loss": 0.2922,
+      "step": 563
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3882986307144165,
+      "learning_rate": 8.597410015797358e-05,
+      "loss": 0.2404,
+      "step": 564
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.448748379945755,
+      "learning_rate": 8.592073736709996e-05,
+      "loss": 0.3581,
+      "step": 565
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.41546115279197693,
+      "learning_rate": 8.586728988593158e-05,
+      "loss": 0.2935,
+      "step": 566
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.46114858984947205,
+      "learning_rate": 8.581375784048153e-05,
+      "loss": 0.3297,
+      "step": 567
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.42799898982048035,
+      "learning_rate": 8.576014135696226e-05,
+      "loss": 0.2936,
+      "step": 568
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3962346613407135,
+      "learning_rate": 8.570644056178532e-05,
+      "loss": 0.2611,
+      "step": 569
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4751885235309601,
+      "learning_rate": 8.565265558156101e-05,
+      "loss": 0.3536,
+      "step": 570
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5154991745948792,
+      "learning_rate": 8.559878654309819e-05,
+      "loss": 0.3738,
+      "step": 571
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4970434308052063,
+      "learning_rate": 8.554483357340379e-05,
+      "loss": 0.3365,
+      "step": 572
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4535207748413086,
+      "learning_rate": 8.549079679968272e-05,
+      "loss": 0.2775,
+      "step": 573
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5101510286331177,
+      "learning_rate": 8.543667634933742e-05,
+      "loss": 0.3626,
+      "step": 574
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.39760786294937134,
+      "learning_rate": 8.538247234996766e-05,
+      "loss": 0.1959,
+      "step": 575
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.466314435005188,
+      "learning_rate": 8.532818492937014e-05,
+      "loss": 0.3197,
+      "step": 576
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.40225160121917725,
+      "learning_rate": 8.527381421553829e-05,
+      "loss": 0.2129,
+      "step": 577
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.44222185015678406,
+      "learning_rate": 8.521936033666187e-05,
+      "loss": 0.2507,
+      "step": 578
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.48492375016212463,
+      "learning_rate": 8.516482342112679e-05,
+      "loss": 0.2792,
+      "step": 579
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.45973947644233704,
+      "learning_rate": 8.511020359751466e-05,
+      "loss": 0.2652,
+      "step": 580
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4101805090904236,
+      "learning_rate": 8.505550099460265e-05,
+      "loss": 0.2239,
+      "step": 581
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4081605076789856,
+      "learning_rate": 8.500071574136295e-05,
+      "loss": 0.2355,
+      "step": 582
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4293803870677948,
+      "learning_rate": 8.49458479669628e-05,
+      "loss": 0.2772,
+      "step": 583
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.40992575883865356,
+      "learning_rate": 8.489089780076386e-05,
+      "loss": 0.2396,
+      "step": 584
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5021704435348511,
+      "learning_rate": 8.483586537232211e-05,
+      "loss": 0.2858,
+      "step": 585
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.44911572337150574,
+      "learning_rate": 8.478075081138745e-05,
+      "loss": 0.2154,
+      "step": 586
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4820658266544342,
+      "learning_rate": 8.472555424790347e-05,
+      "loss": 0.3416,
+      "step": 587
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4892919361591339,
+      "learning_rate": 8.467027581200702e-05,
+      "loss": 0.3435,
+      "step": 588
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5656145811080933,
+      "learning_rate": 8.461491563402807e-05,
+      "loss": 0.3383,
+      "step": 589
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.43777012825012207,
+      "learning_rate": 8.455947384448924e-05,
+      "loss": 0.2912,
+      "step": 590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4463053345680237,
+      "learning_rate": 8.450395057410561e-05,
+      "loss": 0.3142,
+      "step": 591
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.43795090913772583,
+      "learning_rate": 8.444834595378434e-05,
+      "loss": 0.3062,
+      "step": 592
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3950531780719757,
+      "learning_rate": 8.43926601146244e-05,
+      "loss": 0.2133,
+      "step": 593
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5000033378601074,
+      "learning_rate": 8.433689318791626e-05,
+      "loss": 0.3501,
+      "step": 594
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.42796748876571655,
+      "learning_rate": 8.428104530514154e-05,
+      "loss": 0.3097,
+      "step": 595
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38926056027412415,
+      "learning_rate": 8.422511659797279e-05,
+      "loss": 0.2673,
+      "step": 596
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42408907413482666,
+      "learning_rate": 8.416910719827304e-05,
+      "loss": 0.2423,
+      "step": 597
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4298848807811737,
+      "learning_rate": 8.411301723809563e-05,
+      "loss": 0.3149,
+      "step": 598
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4609076976776123,
+      "learning_rate": 8.405684684968383e-05,
+      "loss": 0.2532,
+      "step": 599
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.414275199174881,
+      "learning_rate": 8.400059616547047e-05,
+      "loss": 0.2847,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.28477364778518677,
+      "eval_runtime": 195.5772,
+      "eval_samples_per_second": 1.744,
+      "eval_steps_per_second": 0.22,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4724257290363312,
+      "learning_rate": 8.394426531807778e-05,
+      "loss": 0.4015,
+      "step": 601
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42112022638320923,
+      "learning_rate": 8.388785444031695e-05,
+      "loss": 0.2509,
+      "step": 602
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4972168505191803,
+      "learning_rate": 8.383136366518788e-05,
+      "loss": 0.3047,
+      "step": 603
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.45996132493019104,
+      "learning_rate": 8.377479312587879e-05,
+      "loss": 0.3383,
+      "step": 604
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3250066637992859,
+      "learning_rate": 8.371814295576603e-05,
+      "loss": 0.1577,
+      "step": 605
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.47847071290016174,
+      "learning_rate": 8.366141328841366e-05,
+      "loss": 0.2899,
+      "step": 606
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.49923449754714966,
+      "learning_rate": 8.360460425757314e-05,
+      "loss": 0.3746,
+      "step": 607
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4356243908405304,
+      "learning_rate": 8.354771599718312e-05,
+      "loss": 0.2556,
+      "step": 608
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4790905714035034,
+      "learning_rate": 8.349074864136898e-05,
+      "loss": 0.329,
+      "step": 609
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39675959944725037,
+      "learning_rate": 8.343370232444261e-05,
+      "loss": 0.2646,
+      "step": 610
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5098634958267212,
+      "learning_rate": 8.33765771809021e-05,
+      "loss": 0.3593,
+      "step": 611
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.46119043231010437,
+      "learning_rate": 8.331937334543132e-05,
+      "loss": 0.2839,
+      "step": 612
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4549422264099121,
+      "learning_rate": 8.326209095289972e-05,
+      "loss": 0.2898,
+      "step": 613
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4243537485599518,
+      "learning_rate": 8.320473013836196e-05,
+      "loss": 0.275,
+      "step": 614
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4064863920211792,
+      "learning_rate": 8.314729103705758e-05,
+      "loss": 0.2791,
+      "step": 615
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5189734101295471,
+      "learning_rate": 8.308977378441072e-05,
+      "loss": 0.2812,
+      "step": 616
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3958117663860321,
+      "learning_rate": 8.303217851602973e-05,
+      "loss": 0.2187,
+      "step": 617
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.42362168431282043,
+      "learning_rate": 8.297450536770697e-05,
+      "loss": 0.277,
+      "step": 618
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4793214797973633,
+      "learning_rate": 8.291675447541833e-05,
+      "loss": 0.3412,
+      "step": 619
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.39702266454696655,
+      "learning_rate": 8.28589259753231e-05,
+      "loss": 0.3071,
+      "step": 620
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.47633904218673706,
+      "learning_rate": 8.280102000376345e-05,
+      "loss": 0.2887,
+      "step": 621
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.410567045211792,
+      "learning_rate": 8.274303669726426e-05,
+      "loss": 0.2696,
+      "step": 622
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4697405993938446,
+      "learning_rate": 8.268497619253273e-05,
+      "loss": 0.331,
+      "step": 623
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.40318113565444946,
+      "learning_rate": 8.262683862645804e-05,
+      "loss": 0.2235,
+      "step": 624
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.37693607807159424,
+      "learning_rate": 8.256862413611113e-05,
+      "loss": 0.2432,
+      "step": 625
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.46740347146987915,
+      "learning_rate": 8.251033285874421e-05,
+      "loss": 0.3368,
+      "step": 626
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4473685026168823,
+      "learning_rate": 8.245196493179061e-05,
+      "loss": 0.3268,
+      "step": 627
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4492732286453247,
+      "learning_rate": 8.239352049286435e-05,
+      "loss": 0.3221,
+      "step": 628
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.43746212124824524,
+      "learning_rate": 8.233499967975982e-05,
+      "loss": 0.3626,
+      "step": 629
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3529077172279358,
+      "learning_rate": 8.22764026304515e-05,
+      "loss": 0.2087,
+      "step": 630
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3371795117855072,
+      "learning_rate": 8.221772948309362e-05,
+      "loss": 0.1947,
+      "step": 631
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5318558812141418,
+      "learning_rate": 8.21589803760198e-05,
+      "loss": 0.405,
+      "step": 632
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5322273373603821,
+      "learning_rate": 8.210015544774278e-05,
+      "loss": 0.4448,
+      "step": 633
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.42336592078208923,
+      "learning_rate": 8.204125483695403e-05,
+      "loss": 0.2642,
+      "step": 634
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5167324542999268,
+      "learning_rate": 8.198227868252348e-05,
+      "loss": 0.3952,
+      "step": 635
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4612956941127777,
+      "learning_rate": 8.192322712349917e-05,
+      "loss": 0.3177,
+      "step": 636
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5338868498802185,
+      "learning_rate": 8.186410029910694e-05,
+      "loss": 0.2352,
+      "step": 637
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.44670209288597107,
+      "learning_rate": 8.180489834875e-05,
+      "loss": 0.2877,
+      "step": 638
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4281480014324188,
+      "learning_rate": 8.174562141200878e-05,
+      "loss": 0.3239,
+      "step": 639
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.44553032517433167,
+      "learning_rate": 8.168626962864045e-05,
+      "loss": 0.2491,
+      "step": 640
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.47550174593925476,
+      "learning_rate": 8.16268431385787e-05,
+      "loss": 0.3265,
+      "step": 641
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.43475252389907837,
+      "learning_rate": 8.156734208193327e-05,
+      "loss": 0.3319,
+      "step": 642
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.39503952860832214,
+      "learning_rate": 8.15077665989898e-05,
+      "loss": 0.2578,
+      "step": 643
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.42589524388313293,
+      "learning_rate": 8.144811683020932e-05,
+      "loss": 0.2431,
+      "step": 644
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4075135290622711,
+      "learning_rate": 8.138839291622808e-05,
+      "loss": 0.2635,
+      "step": 645
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5000337362289429,
+      "learning_rate": 8.132859499785707e-05,
+      "loss": 0.363,
+      "step": 646
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.40438759326934814,
+      "learning_rate": 8.126872321608184e-05,
+      "loss": 0.2244,
+      "step": 647
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.36792120337486267,
+      "learning_rate": 8.120877771206201e-05,
+      "loss": 0.2264,
+      "step": 648
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4347400665283203,
+      "learning_rate": 8.114875862713106e-05,
+      "loss": 0.3017,
+      "step": 649
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4728121757507324,
+      "learning_rate": 8.108866610279593e-05,
+      "loss": 0.3172,
+      "step": 650
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4600279927253723,
+      "learning_rate": 8.102850028073674e-05,
+      "loss": 0.2981,
+      "step": 651
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4729394316673279,
+      "learning_rate": 8.096826130280639e-05,
+      "loss": 0.3498,
+      "step": 652
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36024209856987,
+      "learning_rate": 8.090794931103026e-05,
+      "loss": 0.2715,
+      "step": 653
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.37017837166786194,
+      "learning_rate": 8.08475644476059e-05,
+      "loss": 0.2103,
+      "step": 654
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4497227668762207,
+      "learning_rate": 8.078710685490265e-05,
+      "loss": 0.3456,
+      "step": 655
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4679158627986908,
+      "learning_rate": 8.072657667546136e-05,
+      "loss": 0.2301,
+      "step": 656
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5199421644210815,
+      "learning_rate": 8.066597405199392e-05,
+      "loss": 0.3243,
+      "step": 657
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3976512551307678,
+      "learning_rate": 8.060529912738315e-05,
+      "loss": 0.24,
+      "step": 658
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.506020188331604,
+      "learning_rate": 8.054455204468224e-05,
+      "loss": 0.3423,
+      "step": 659
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.479912132024765,
+      "learning_rate": 8.048373294711455e-05,
+      "loss": 0.3404,
+      "step": 660
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4801808297634125,
+      "learning_rate": 8.042284197807324e-05,
+      "loss": 0.2791,
+      "step": 661
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4489070177078247,
+      "learning_rate": 8.036187928112087e-05,
+      "loss": 0.3066,
+      "step": 662
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4340127408504486,
+      "learning_rate": 8.030084499998916e-05,
+      "loss": 0.2867,
+      "step": 663
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4398484230041504,
+      "learning_rate": 8.023973927857857e-05,
+      "loss": 0.3418,
+      "step": 664
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4984659254550934,
+      "learning_rate": 8.017856226095804e-05,
+      "loss": 0.3546,
+      "step": 665
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.43073058128356934,
+      "learning_rate": 8.011731409136454e-05,
+      "loss": 0.2907,
+      "step": 666
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.42610689997673035,
+      "learning_rate": 8.005599491420288e-05,
+      "loss": 0.3111,
+      "step": 667
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4633261561393738,
+      "learning_rate": 7.99946048740452e-05,
+      "loss": 0.3148,
+      "step": 668
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3821967840194702,
+      "learning_rate": 7.993314411563075e-05,
+      "loss": 0.239,
+      "step": 669
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.39985767006874084,
+      "learning_rate": 7.987161278386554e-05,
+      "loss": 0.2595,
+      "step": 670
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5795412063598633,
+      "learning_rate": 7.981001102382192e-05,
+      "loss": 0.3999,
+      "step": 671
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.42855364084243774,
+      "learning_rate": 7.974833898073832e-05,
+      "loss": 0.3009,
+      "step": 672
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4170435965061188,
+      "learning_rate": 7.968659680001886e-05,
+      "loss": 0.289,
+      "step": 673
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.42596131563186646,
+      "learning_rate": 7.962478462723307e-05,
+      "loss": 0.2762,
+      "step": 674
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4302665591239929,
+      "learning_rate": 7.95629026081154e-05,
+      "loss": 0.2692,
+      "step": 675
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4310140013694763,
+      "learning_rate": 7.950095088856508e-05,
+      "loss": 0.3253,
+      "step": 676
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.378837525844574,
+      "learning_rate": 7.943892961464563e-05,
+      "loss": 0.1991,
+      "step": 677
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.43026968836784363,
+      "learning_rate": 7.937683893258454e-05,
+      "loss": 0.2669,
+      "step": 678
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4046042263507843,
+      "learning_rate": 7.931467898877297e-05,
+      "loss": 0.28,
+      "step": 679
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.38268038630485535,
+      "learning_rate": 7.925244992976538e-05,
+      "loss": 0.2974,
+      "step": 680
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.43971017003059387,
+      "learning_rate": 7.919015190227919e-05,
+      "loss": 0.3259,
+      "step": 681
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.479377806186676,
+      "learning_rate": 7.912778505319436e-05,
+      "loss": 0.2456,
+      "step": 682
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.47326013445854187,
+      "learning_rate": 7.906534952955321e-05,
+      "loss": 0.2932,
+      "step": 683
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5107829570770264,
+      "learning_rate": 7.900284547855991e-05,
+      "loss": 0.4094,
+      "step": 684
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.48692771792411804,
+      "learning_rate": 7.894027304758022e-05,
+      "loss": 0.3152,
+      "step": 685
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.47162333130836487,
+      "learning_rate": 7.88776323841411e-05,
+      "loss": 0.3098,
+      "step": 686
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.46147269010543823,
+      "learning_rate": 7.881492363593039e-05,
+      "loss": 0.3162,
+      "step": 687
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.47289565205574036,
+      "learning_rate": 7.875214695079647e-05,
+      "loss": 0.3029,
+      "step": 688
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.40165475010871887,
+      "learning_rate": 7.868930247674788e-05,
+      "loss": 0.2664,
+      "step": 689
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4063616394996643,
+      "learning_rate": 7.862639036195298e-05,
+      "loss": 0.2192,
+      "step": 690
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.44326379895210266,
+      "learning_rate": 7.856341075473962e-05,
+      "loss": 0.3078,
+      "step": 691
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.550214409828186,
+      "learning_rate": 7.850036380359478e-05,
+      "loss": 0.3987,
+      "step": 692
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.49500808119773865,
+      "learning_rate": 7.843724965716419e-05,
+      "loss": 0.3553,
+      "step": 693
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4465545117855072,
+      "learning_rate": 7.837406846425204e-05,
+      "loss": 0.2171,
+      "step": 694
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5085720419883728,
+      "learning_rate": 7.831082037382057e-05,
+      "loss": 0.3366,
+      "step": 695
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3921324908733368,
+      "learning_rate": 7.824750553498976e-05,
+      "loss": 0.2626,
+      "step": 696
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.44672077894210815,
+      "learning_rate": 7.818412409703695e-05,
+      "loss": 0.3021,
+      "step": 697
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.37329670786857605,
+      "learning_rate": 7.812067620939652e-05,
+      "loss": 0.2468,
+      "step": 698
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.46521878242492676,
+      "learning_rate": 7.805716202165949e-05,
+      "loss": 0.3964,
+      "step": 699
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.46653467416763306,
+      "learning_rate": 7.799358168357323e-05,
+      "loss": 0.3337,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.28784823417663574,
+      "eval_runtime": 195.104,
+      "eval_samples_per_second": 1.748,
+      "eval_steps_per_second": 0.22,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4581441581249237,
+      "learning_rate": 7.792993534504102e-05,
+      "loss": 0.3092,
+      "step": 701
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4534381628036499,
+      "learning_rate": 7.786622315612183e-05,
+      "loss": 0.3474,
+      "step": 702
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3672908842563629,
+      "learning_rate": 7.78024452670298e-05,
+      "loss": 0.2153,
+      "step": 703
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5289280414581299,
+      "learning_rate": 7.773860182813403e-05,
+      "loss": 0.3824,
+      "step": 704
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3858056366443634,
+      "learning_rate": 7.767469298995813e-05,
+      "loss": 0.2676,
+      "step": 705
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.445822536945343,
+      "learning_rate": 7.761071890317994e-05,
+      "loss": 0.3171,
+      "step": 706
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40967124700546265,
+      "learning_rate": 7.754667971863112e-05,
+      "loss": 0.2787,
+      "step": 707
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3603118658065796,
+      "learning_rate": 7.748257558729677e-05,
+      "loss": 0.2108,
+      "step": 708
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.40177345275878906,
+      "learning_rate": 7.741840666031517e-05,
+      "loss": 0.2195,
+      "step": 709
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4547106921672821,
+      "learning_rate": 7.735417308897736e-05,
+      "loss": 0.2904,
+      "step": 710
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.43407630920410156,
+      "learning_rate": 7.728987502472677e-05,
+      "loss": 0.3196,
+      "step": 711
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4073294699192047,
+      "learning_rate": 7.72255126191589e-05,
+      "loss": 0.2259,
+      "step": 712
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.41981470584869385,
+      "learning_rate": 7.716108602402094e-05,
+      "loss": 0.2803,
+      "step": 713
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5095283389091492,
+      "learning_rate": 7.709659539121144e-05,
+      "loss": 0.3814,
+      "step": 714
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4663334786891937,
+      "learning_rate": 7.703204087277988e-05,
+      "loss": 0.2977,
+      "step": 715
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4155025780200958,
+      "learning_rate": 7.696742262092644e-05,
+      "loss": 0.2763,
+      "step": 716
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4163815975189209,
+      "learning_rate": 7.690274078800149e-05,
+      "loss": 0.2389,
+      "step": 717
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.40252846479415894,
+      "learning_rate": 7.683799552650534e-05,
+      "loss": 0.2433,
+      "step": 718
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4456719756126404,
+      "learning_rate": 7.677318698908789e-05,
+      "loss": 0.2646,
+      "step": 719
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5050420165061951,
+      "learning_rate": 7.67083153285481e-05,
+      "loss": 0.38,
+      "step": 720
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4679016172885895,
+      "learning_rate": 7.66433806978339e-05,
+      "loss": 0.3325,
+      "step": 721
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4437169134616852,
+      "learning_rate": 7.657838325004159e-05,
+      "loss": 0.3469,
+      "step": 722
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.38237640261650085,
+      "learning_rate": 7.651332313841562e-05,
+      "loss": 0.2224,
+      "step": 723
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5386263132095337,
+      "learning_rate": 7.644820051634812e-05,
+      "loss": 0.3542,
+      "step": 724
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.430181622505188,
+      "learning_rate": 7.638301553737869e-05,
+      "loss": 0.2638,
+      "step": 725
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4215375483036041,
+      "learning_rate": 7.63177683551939e-05,
+      "loss": 0.2102,
+      "step": 726
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3984241485595703,
+      "learning_rate": 7.625245912362699e-05,
+      "loss": 0.2147,
+      "step": 727
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3855826258659363,
+      "learning_rate": 7.618708799665744e-05,
+      "loss": 0.2374,
+      "step": 728
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.44361966848373413,
+      "learning_rate": 7.612165512841075e-05,
+      "loss": 0.2759,
+      "step": 729
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5300565361976624,
+      "learning_rate": 7.605616067315793e-05,
+      "loss": 0.3349,
+      "step": 730
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.40108662843704224,
+      "learning_rate": 7.59906047853152e-05,
+      "loss": 0.2437,
+      "step": 731
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4717155992984772,
+      "learning_rate": 7.592498761944362e-05,
+      "loss": 0.2469,
+      "step": 732
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4769246578216553,
+      "learning_rate": 7.585930933024875e-05,
+      "loss": 0.2743,
+      "step": 733
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5374858379364014,
+      "learning_rate": 7.579357007258022e-05,
+      "loss": 0.3802,
+      "step": 734
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5996991991996765,
+      "learning_rate": 7.572777000143145e-05,
+      "loss": 0.3403,
+      "step": 735
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.549309492111206,
+      "learning_rate": 7.56619092719392e-05,
+      "loss": 0.3939,
+      "step": 736
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3180038034915924,
+      "learning_rate": 7.559598803938328e-05,
+      "loss": 0.1809,
+      "step": 737
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4131849408149719,
+      "learning_rate": 7.553000645918611e-05,
+      "loss": 0.2462,
+      "step": 738
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4472147226333618,
+      "learning_rate": 7.546396468691242e-05,
+      "loss": 0.2882,
+      "step": 739
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4557483196258545,
+      "learning_rate": 7.539786287826885e-05,
+      "loss": 0.3258,
+      "step": 740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.39649051427841187,
+      "learning_rate": 7.533170118910355e-05,
+      "loss": 0.2345,
+      "step": 741
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4275149405002594,
+      "learning_rate": 7.526547977540592e-05,
+      "loss": 0.2939,
+      "step": 742
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3712472915649414,
+      "learning_rate": 7.519919879330614e-05,
+      "loss": 0.2459,
+      "step": 743
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.41389191150665283,
+      "learning_rate": 7.51328583990748e-05,
+      "loss": 0.2837,
+      "step": 744
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.42736154794692993,
+      "learning_rate": 7.506645874912264e-05,
+      "loss": 0.311,
+      "step": 745
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.42053690552711487,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.2736,
+      "step": 746
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.39817917346954346,
+      "learning_rate": 7.493348230839668e-05,
+      "loss": 0.1815,
+      "step": 747
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5271608233451843,
+      "learning_rate": 7.486690583114136e-05,
+      "loss": 0.3652,
+      "step": 748
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.44614550471305847,
+      "learning_rate": 7.480027072520137e-05,
+      "loss": 0.3032,
+      "step": 749
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3886353075504303,
+      "learning_rate": 7.473357714768222e-05,
+      "loss": 0.2343,
+      "step": 750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.41741499304771423,
+      "learning_rate": 7.46668252558273e-05,
+      "loss": 0.2584,
+      "step": 751
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4288809597492218,
+      "learning_rate": 7.460001520701755e-05,
+      "loss": 0.3025,
+      "step": 752
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.37444329261779785,
+      "learning_rate": 7.453314715877093e-05,
+      "loss": 0.2376,
+      "step": 753
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.43826186656951904,
+      "learning_rate": 7.446622126874218e-05,
+      "loss": 0.2394,
+      "step": 754
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.502704918384552,
+      "learning_rate": 7.439923769472243e-05,
+      "loss": 0.3606,
+      "step": 755
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4681566059589386,
+      "learning_rate": 7.433219659463881e-05,
+      "loss": 0.3843,
+      "step": 756
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.39780882000923157,
+      "learning_rate": 7.426509812655406e-05,
+      "loss": 0.2349,
+      "step": 757
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4249710738658905,
+      "learning_rate": 7.41979424486662e-05,
+      "loss": 0.2919,
+      "step": 758
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.528399646282196,
+      "learning_rate": 7.413072971930807e-05,
+      "loss": 0.3961,
+      "step": 759
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5760734677314758,
+      "learning_rate": 7.406346009694713e-05,
+      "loss": 0.4808,
+      "step": 760
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3934260904788971,
+      "learning_rate": 7.399613374018489e-05,
+      "loss": 0.2932,
+      "step": 761
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.45331597328186035,
+      "learning_rate": 7.392875080775665e-05,
+      "loss": 0.2407,
+      "step": 762
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.41213762760162354,
+      "learning_rate": 7.386131145853112e-05,
+      "loss": 0.2858,
+      "step": 763
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.41022974252700806,
+      "learning_rate": 7.379381585150996e-05,
+      "loss": 0.2441,
+      "step": 764
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4302038252353668,
+      "learning_rate": 7.372626414582754e-05,
+      "loss": 0.2934,
+      "step": 765
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5060116052627563,
+      "learning_rate": 7.365865650075046e-05,
+      "loss": 0.3039,
+      "step": 766
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4041798412799835,
+      "learning_rate": 7.359099307567721e-05,
+      "loss": 0.2197,
+      "step": 767
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.45071354508399963,
+      "learning_rate": 7.35232740301378e-05,
+      "loss": 0.3288,
+      "step": 768
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.48885127902030945,
+      "learning_rate": 7.345549952379334e-05,
+      "loss": 0.331,
+      "step": 769
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3668765425682068,
+      "learning_rate": 7.338766971643579e-05,
+      "loss": 0.2494,
+      "step": 770
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4049915671348572,
+      "learning_rate": 7.331978476798738e-05,
+      "loss": 0.2857,
+      "step": 771
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.48406821489334106,
+      "learning_rate": 7.325184483850042e-05,
+      "loss": 0.3328,
+      "step": 772
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.500324547290802,
+      "learning_rate": 7.318385008815686e-05,
+      "loss": 0.3563,
+      "step": 773
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.43897345662117004,
+      "learning_rate": 7.311580067726783e-05,
+      "loss": 0.293,
+      "step": 774
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4582681953907013,
+      "learning_rate": 7.304769676627339e-05,
+      "loss": 0.3696,
+      "step": 775
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2897372245788574,
+      "learning_rate": 7.297953851574206e-05,
+      "loss": 0.1631,
+      "step": 776
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.48981499671936035,
+      "learning_rate": 7.291132608637052e-05,
+      "loss": 0.3816,
+      "step": 777
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4538932144641876,
+      "learning_rate": 7.284305963898314e-05,
+      "loss": 0.3025,
+      "step": 778
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.33093053102493286,
+      "learning_rate": 7.27747393345317e-05,
+      "loss": 0.1797,
+      "step": 779
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.45261475443840027,
+      "learning_rate": 7.27063653340949e-05,
+      "loss": 0.3158,
+      "step": 780
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4943481683731079,
+      "learning_rate": 7.263793779887808e-05,
+      "loss": 0.3264,
+      "step": 781
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.9289783239364624,
+      "learning_rate": 7.25694568902128e-05,
+      "loss": 0.3756,
+      "step": 782
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3444068133831024,
+      "learning_rate": 7.250092276955642e-05,
+      "loss": 0.2181,
+      "step": 783
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.47700634598731995,
+      "learning_rate": 7.243233559849179e-05,
+      "loss": 0.3513,
+      "step": 784
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41721591353416443,
+      "learning_rate": 7.236369553872685e-05,
+      "loss": 0.3334,
+      "step": 785
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4001460671424866,
+      "learning_rate": 7.229500275209418e-05,
+      "loss": 0.2935,
+      "step": 786
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.43575119972229004,
+      "learning_rate": 7.222625740055073e-05,
+      "loss": 0.2726,
+      "step": 787
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.44586196541786194,
+      "learning_rate": 7.215745964617737e-05,
+      "loss": 0.2904,
+      "step": 788
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.39179566502571106,
+      "learning_rate": 7.20886096511785e-05,
+      "loss": 0.241,
+      "step": 789
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3982108533382416,
+      "learning_rate": 7.201970757788172e-05,
+      "loss": 0.2784,
+      "step": 790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.36585283279418945,
+      "learning_rate": 7.195075358873738e-05,
+      "loss": 0.2346,
+      "step": 791
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.42613306641578674,
+      "learning_rate": 7.188174784631824e-05,
+      "loss": 0.2974,
+      "step": 792
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.40931862592697144,
+      "learning_rate": 7.181269051331909e-05,
+      "loss": 0.2542,
+      "step": 793
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4294385313987732,
+      "learning_rate": 7.174358175255636e-05,
+      "loss": 0.3136,
+      "step": 794
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.44175368547439575,
+      "learning_rate": 7.16744217269677e-05,
+      "loss": 0.3566,
+      "step": 795
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4859725534915924,
+      "learning_rate": 7.160521059961169e-05,
+      "loss": 0.2964,
+      "step": 796
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.43112170696258545,
+      "learning_rate": 7.153594853366731e-05,
+      "loss": 0.2924,
+      "step": 797
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.45823830366134644,
+      "learning_rate": 7.146663569243371e-05,
+      "loss": 0.2932,
+      "step": 798
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.46234431862831116,
+      "learning_rate": 7.139727223932968e-05,
+      "loss": 0.3605,
+      "step": 799
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3390848636627197,
+      "learning_rate": 7.132785833789344e-05,
+      "loss": 0.1648,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.28161272406578064,
+      "eval_runtime": 194.896,
+      "eval_samples_per_second": 1.75,
+      "eval_steps_per_second": 0.221,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5338891744613647,
+      "learning_rate": 7.125839415178204e-05,
+      "loss": 0.45,
+      "step": 801
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3764931559562683,
+      "learning_rate": 7.118887984477116e-05,
+      "loss": 0.2024,
+      "step": 802
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.46116527915000916,
+      "learning_rate": 7.111931558075464e-05,
+      "loss": 0.3339,
+      "step": 803
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4544364809989929,
+      "learning_rate": 7.104970152374405e-05,
+      "loss": 0.2481,
+      "step": 804
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5027008056640625,
+      "learning_rate": 7.098003783786845e-05,
+      "loss": 0.3548,
+      "step": 805
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.450987309217453,
+      "learning_rate": 7.091032468737381e-05,
+      "loss": 0.2595,
+      "step": 806
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.43323996663093567,
+      "learning_rate": 7.084056223662282e-05,
+      "loss": 0.3003,
+      "step": 807
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4585069715976715,
+      "learning_rate": 7.077075065009433e-05,
+      "loss": 0.3248,
+      "step": 808
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4782935678958893,
+      "learning_rate": 7.070089009238306e-05,
+      "loss": 0.323,
+      "step": 809
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.48253458738327026,
+      "learning_rate": 7.063098072819919e-05,
+      "loss": 0.3281,
+      "step": 810
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3753528594970703,
+      "learning_rate": 7.056102272236799e-05,
+      "loss": 0.2163,
+      "step": 811
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5140417814254761,
+      "learning_rate": 7.049101623982937e-05,
+      "loss": 0.3057,
+      "step": 812
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.43106040358543396,
+      "learning_rate": 7.04209614456376e-05,
+      "loss": 0.2801,
+      "step": 813
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4178459048271179,
+      "learning_rate": 7.03508585049608e-05,
+      "loss": 0.2863,
+      "step": 814
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3956921100616455,
+      "learning_rate": 7.028070758308059e-05,
+      "loss": 0.2421,
+      "step": 815
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4430694878101349,
+      "learning_rate": 7.021050884539177e-05,
+      "loss": 0.2965,
+      "step": 816
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.42406195402145386,
+      "learning_rate": 7.014026245740186e-05,
+      "loss": 0.261,
+      "step": 817
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5281689167022705,
+      "learning_rate": 7.006996858473069e-05,
+      "loss": 0.3714,
+      "step": 818
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.429524302482605,
+      "learning_rate": 6.999962739311008e-05,
+      "loss": 0.2335,
+      "step": 819
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5070218443870544,
+      "learning_rate": 6.99292390483834e-05,
+      "loss": 0.3672,
+      "step": 820
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6425400376319885,
+      "learning_rate": 6.98588037165052e-05,
+      "loss": 0.3185,
+      "step": 821
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.46291404962539673,
+      "learning_rate": 6.97883215635408e-05,
+      "loss": 0.3248,
+      "step": 822
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4308122396469116,
+      "learning_rate": 6.971779275566593e-05,
+      "loss": 0.3183,
+      "step": 823
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.34428325295448303,
+      "learning_rate": 6.964721745916631e-05,
+      "loss": 0.2282,
+      "step": 824
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.44850701093673706,
+      "learning_rate": 6.957659584043724e-05,
+      "loss": 0.2983,
+      "step": 825
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3743911683559418,
+      "learning_rate": 6.950592806598327e-05,
+      "loss": 0.2705,
+      "step": 826
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4245792031288147,
+      "learning_rate": 6.943521430241776e-05,
+      "loss": 0.3171,
+      "step": 827
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4242267310619354,
+      "learning_rate": 6.936445471646249e-05,
+      "loss": 0.2949,
+      "step": 828
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4110967516899109,
+      "learning_rate": 6.929364947494728e-05,
+      "loss": 0.3059,
+      "step": 829
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.39392754435539246,
+      "learning_rate": 6.922279874480959e-05,
+      "loss": 0.3034,
+      "step": 830
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.42310869693756104,
+      "learning_rate": 6.915190269309414e-05,
+      "loss": 0.3473,
+      "step": 831
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.416714072227478,
+      "learning_rate": 6.908096148695251e-05,
+      "loss": 0.2899,
+      "step": 832
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4073675572872162,
+      "learning_rate": 6.90099752936427e-05,
+      "loss": 0.2205,
+      "step": 833
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4612197279930115,
+      "learning_rate": 6.89389442805288e-05,
+      "loss": 0.3714,
+      "step": 834
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.38989490270614624,
+      "learning_rate": 6.886786861508061e-05,
+      "loss": 0.2554,
+      "step": 835
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4297979176044464,
+      "learning_rate": 6.879674846487314e-05,
+      "loss": 0.3626,
+      "step": 836
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3973498046398163,
+      "learning_rate": 6.872558399758632e-05,
+      "loss": 0.3014,
+      "step": 837
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3641144335269928,
+      "learning_rate": 6.865437538100457e-05,
+      "loss": 0.1778,
+      "step": 838
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.40729624032974243,
+      "learning_rate": 6.858312278301637e-05,
+      "loss": 0.2867,
+      "step": 839
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.45078954100608826,
+      "learning_rate": 6.851182637161396e-05,
+      "loss": 0.3919,
+      "step": 840
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38180261850357056,
+      "learning_rate": 6.844048631489278e-05,
+      "loss": 0.2348,
+      "step": 841
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4587530195713043,
+      "learning_rate": 6.836910278105124e-05,
+      "loss": 0.3484,
+      "step": 842
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4379264712333679,
+      "learning_rate": 6.829767593839028e-05,
+      "loss": 0.3076,
+      "step": 843
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3916952610015869,
+      "learning_rate": 6.822620595531286e-05,
+      "loss": 0.1844,
+      "step": 844
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.35218703746795654,
+      "learning_rate": 6.815469300032374e-05,
+      "loss": 0.2142,
+      "step": 845
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.46940281987190247,
+      "learning_rate": 6.808313724202894e-05,
+      "loss": 0.3652,
+      "step": 846
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4841424822807312,
+      "learning_rate": 6.801153884913542e-05,
+      "loss": 0.396,
+      "step": 847
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.40121644735336304,
+      "learning_rate": 6.793989799045066e-05,
+      "loss": 0.2818,
+      "step": 848
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.42353928089141846,
+      "learning_rate": 6.786821483488226e-05,
+      "loss": 0.2678,
+      "step": 849
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.373620867729187,
+      "learning_rate": 6.779648955143754e-05,
+      "loss": 0.1856,
+      "step": 850
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.44443953037261963,
+      "learning_rate": 6.772472230922313e-05,
+      "loss": 0.3101,
+      "step": 851
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.36101463437080383,
+      "learning_rate": 6.765291327744463e-05,
+      "loss": 0.2398,
+      "step": 852
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4875366687774658,
+      "learning_rate": 6.758106262540611e-05,
+      "loss": 0.2766,
+      "step": 853
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.42991340160369873,
+      "learning_rate": 6.750917052250982e-05,
+      "loss": 0.2774,
+      "step": 854
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.35289445519447327,
+      "learning_rate": 6.743723713825569e-05,
+      "loss": 0.1985,
+      "step": 855
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3485260009765625,
+      "learning_rate": 6.736526264224101e-05,
+      "loss": 0.1844,
+      "step": 856
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3996266722679138,
+      "learning_rate": 6.729324720416002e-05,
+      "loss": 0.2453,
+      "step": 857
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.43274182081222534,
+      "learning_rate": 6.722119099380345e-05,
+      "loss": 0.2861,
+      "step": 858
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.35542285442352295,
+      "learning_rate": 6.714909418105815e-05,
+      "loss": 0.1957,
+      "step": 859
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.405661016702652,
+      "learning_rate": 6.707695693590676e-05,
+      "loss": 0.3092,
+      "step": 860
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.47031450271606445,
+      "learning_rate": 6.700477942842717e-05,
+      "loss": 0.2977,
+      "step": 861
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4444357752799988,
+      "learning_rate": 6.693256182879225e-05,
+      "loss": 0.274,
+      "step": 862
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.46726709604263306,
+      "learning_rate": 6.686030430726937e-05,
+      "loss": 0.2995,
+      "step": 863
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.44485777616500854,
+      "learning_rate": 6.678800703422003e-05,
+      "loss": 0.3492,
+      "step": 864
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5059626698493958,
+      "learning_rate": 6.671567018009947e-05,
+      "loss": 0.3696,
+      "step": 865
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4609401226043701,
+      "learning_rate": 6.664329391545626e-05,
+      "loss": 0.2425,
+      "step": 866
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.49468713998794556,
+      "learning_rate": 6.65708784109318e-05,
+      "loss": 0.3073,
+      "step": 867
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.41707897186279297,
+      "learning_rate": 6.64984238372601e-05,
+      "loss": 0.2689,
+      "step": 868
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4629487097263336,
+      "learning_rate": 6.642593036526728e-05,
+      "loss": 0.3215,
+      "step": 869
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4932032525539398,
+      "learning_rate": 6.635339816587109e-05,
+      "loss": 0.3903,
+      "step": 870
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.41509997844696045,
+      "learning_rate": 6.628082741008069e-05,
+      "loss": 0.3092,
+      "step": 871
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5148477554321289,
+      "learning_rate": 6.620821826899606e-05,
+      "loss": 0.3567,
+      "step": 872
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4157521724700928,
+      "learning_rate": 6.613557091380771e-05,
+      "loss": 0.281,
+      "step": 873
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4080047905445099,
+      "learning_rate": 6.606288551579629e-05,
+      "loss": 0.2994,
+      "step": 874
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5310486555099487,
+      "learning_rate": 6.599016224633209e-05,
+      "loss": 0.3201,
+      "step": 875
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4008110761642456,
+      "learning_rate": 6.59174012768747e-05,
+      "loss": 0.29,
+      "step": 876
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38788172602653503,
+      "learning_rate": 6.584460277897263e-05,
+      "loss": 0.2733,
+      "step": 877
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.48622459173202515,
+      "learning_rate": 6.577176692426279e-05,
+      "loss": 0.3449,
+      "step": 878
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.399658739566803,
+      "learning_rate": 6.569889388447025e-05,
+      "loss": 0.3222,
+      "step": 879
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3425334095954895,
+      "learning_rate": 6.562598383140772e-05,
+      "loss": 0.2332,
+      "step": 880
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.395344078540802,
+      "learning_rate": 6.555303693697517e-05,
+      "loss": 0.2336,
+      "step": 881
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4164277911186218,
+      "learning_rate": 6.548005337315944e-05,
+      "loss": 0.2639,
+      "step": 882
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41264939308166504,
+      "learning_rate": 6.540703331203381e-05,
+      "loss": 0.3052,
+      "step": 883
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3729918301105499,
+      "learning_rate": 6.533397692575766e-05,
+      "loss": 0.2214,
+      "step": 884
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4344700872898102,
+      "learning_rate": 6.526088438657594e-05,
+      "loss": 0.2437,
+      "step": 885
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41087377071380615,
+      "learning_rate": 6.518775586681887e-05,
+      "loss": 0.2254,
+      "step": 886
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.45486581325531006,
+      "learning_rate": 6.511459153890152e-05,
+      "loss": 0.2972,
+      "step": 887
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41802269220352173,
+      "learning_rate": 6.504139157532337e-05,
+      "loss": 0.2811,
+      "step": 888
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4393295347690582,
+      "learning_rate": 6.496815614866791e-05,
+      "loss": 0.3097,
+      "step": 889
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.47603145241737366,
+      "learning_rate": 6.489488543160224e-05,
+      "loss": 0.3266,
+      "step": 890
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.48033154010772705,
+      "learning_rate": 6.48215795968767e-05,
+      "loss": 0.2858,
+      "step": 891
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.43115532398223877,
+      "learning_rate": 6.47482388173244e-05,
+      "loss": 0.248,
+      "step": 892
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3620755970478058,
+      "learning_rate": 6.46748632658608e-05,
+      "loss": 0.1709,
+      "step": 893
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.41209709644317627,
+      "learning_rate": 6.460145311548341e-05,
+      "loss": 0.263,
+      "step": 894
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.46709179878234863,
+      "learning_rate": 6.452800853927128e-05,
+      "loss": 0.2847,
+      "step": 895
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.41937920451164246,
+      "learning_rate": 6.445452971038464e-05,
+      "loss": 0.2828,
+      "step": 896
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.511926531791687,
+      "learning_rate": 6.438101680206444e-05,
+      "loss": 0.3534,
+      "step": 897
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.45813050866127014,
+      "learning_rate": 6.430746998763204e-05,
+      "loss": 0.3585,
+      "step": 898
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.45049378275871277,
+      "learning_rate": 6.42338894404887e-05,
+      "loss": 0.3238,
+      "step": 899
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5324050188064575,
+      "learning_rate": 6.416027533411519e-05,
+      "loss": 0.3182,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.2795392870903015,
+      "eval_runtime": 195.6478,
+      "eval_samples_per_second": 1.743,
+      "eval_steps_per_second": 0.22,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.44435036182403564,
+      "learning_rate": 6.408662784207149e-05,
+      "loss": 0.3445,
+      "step": 901
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.43596604466438293,
+      "learning_rate": 6.401294713799618e-05,
+      "loss": 0.3326,
+      "step": 902
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.45252668857574463,
+      "learning_rate": 6.393923339560621e-05,
+      "loss": 0.2738,
+      "step": 903
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4004136323928833,
+      "learning_rate": 6.386548678869644e-05,
+      "loss": 0.2711,
+      "step": 904
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.46836936473846436,
+      "learning_rate": 6.379170749113917e-05,
+      "loss": 0.2655,
+      "step": 905
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.45389804244041443,
+      "learning_rate": 6.37178956768838e-05,
+      "loss": 0.3344,
+      "step": 906
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4013293385505676,
+      "learning_rate": 6.364405151995635e-05,
+      "loss": 0.2929,
+      "step": 907
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4121876060962677,
+      "learning_rate": 6.35701751944592e-05,
+      "loss": 0.2828,
+      "step": 908
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.37906256318092346,
+      "learning_rate": 6.349626687457045e-05,
+      "loss": 0.2322,
+      "step": 909
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4815737009048462,
+      "learning_rate": 6.342232673454371e-05,
+      "loss": 0.3588,
+      "step": 910
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4067385494709015,
+      "learning_rate": 6.334835494870758e-05,
+      "loss": 0.2798,
+      "step": 911
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4159270226955414,
+      "learning_rate": 6.32743516914653e-05,
+      "loss": 0.3332,
+      "step": 912
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4407503008842468,
+      "learning_rate": 6.320031713729428e-05,
+      "loss": 0.2735,
+      "step": 913
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.47542205452919006,
+      "learning_rate": 6.312625146074575e-05,
+      "loss": 0.3876,
+      "step": 914
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.42438551783561707,
+      "learning_rate": 6.305215483644426e-05,
+      "loss": 0.2665,
+      "step": 915
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5114543437957764,
+      "learning_rate": 6.297802743908741e-05,
+      "loss": 0.3558,
+      "step": 916
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4331636130809784,
+      "learning_rate": 6.290386944344528e-05,
+      "loss": 0.3081,
+      "step": 917
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4729527235031128,
+      "learning_rate": 6.282968102436009e-05,
+      "loss": 0.3122,
+      "step": 918
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.40394240617752075,
+      "learning_rate": 6.275546235674588e-05,
+      "loss": 0.3217,
+      "step": 919
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4654287099838257,
+      "learning_rate": 6.268121361558792e-05,
+      "loss": 0.3584,
+      "step": 920
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.42064327001571655,
+      "learning_rate": 6.260693497594242e-05,
+      "loss": 0.2696,
+      "step": 921
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.46156254410743713,
+      "learning_rate": 6.253262661293604e-05,
+      "loss": 0.3788,
+      "step": 922
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.46253877878189087,
+      "learning_rate": 6.245828870176557e-05,
+      "loss": 0.3313,
+      "step": 923
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.38839802145957947,
+      "learning_rate": 6.238392141769743e-05,
+      "loss": 0.2502,
+      "step": 924
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.46288490295410156,
+      "learning_rate": 6.230952493606734e-05,
+      "loss": 0.3422,
+      "step": 925
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4135326147079468,
+      "learning_rate": 6.22350994322798e-05,
+      "loss": 0.253,
+      "step": 926
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.44113191962242126,
+      "learning_rate": 6.216064508180778e-05,
+      "loss": 0.3336,
+      "step": 927
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.39661139249801636,
+      "learning_rate": 6.208616206019224e-05,
+      "loss": 0.2849,
+      "step": 928
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4273655414581299,
+      "learning_rate": 6.201165054304173e-05,
+      "loss": 0.3105,
+      "step": 929
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.459600031375885,
+      "learning_rate": 6.193711070603202e-05,
+      "loss": 0.2551,
+      "step": 930
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4137110114097595,
+      "learning_rate": 6.18625427249056e-05,
+      "loss": 0.293,
+      "step": 931
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4595624506473541,
+      "learning_rate": 6.178794677547137e-05,
+      "loss": 0.339,
+      "step": 932
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4187440574169159,
+      "learning_rate": 6.171332303360411e-05,
+      "loss": 0.3288,
+      "step": 933
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4432730972766876,
+      "learning_rate": 6.163867167524419e-05,
+      "loss": 0.3293,
+      "step": 934
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4180636703968048,
+      "learning_rate": 6.156399287639703e-05,
+      "loss": 0.313,
+      "step": 935
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4032476246356964,
+      "learning_rate": 6.14892868131328e-05,
+      "loss": 0.2353,
+      "step": 936
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.38968080282211304,
+      "learning_rate": 6.14145536615859e-05,
+      "loss": 0.2513,
+      "step": 937
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3711206018924713,
+      "learning_rate": 6.133979359795468e-05,
+      "loss": 0.1922,
+      "step": 938
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4283756613731384,
+      "learning_rate": 6.126500679850083e-05,
+      "loss": 0.3541,
+      "step": 939
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.43274402618408203,
+      "learning_rate": 6.119019343954914e-05,
+      "loss": 0.3053,
+      "step": 940
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.46821972727775574,
+      "learning_rate": 6.111535369748702e-05,
+      "loss": 0.3019,
+      "step": 941
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4059254825115204,
+      "learning_rate": 6.104048774876406e-05,
+      "loss": 0.282,
+      "step": 942
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.420236736536026,
+      "learning_rate": 6.096559576989166e-05,
+      "loss": 0.2798,
+      "step": 943
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3970169723033905,
+      "learning_rate": 6.0890677937442574e-05,
+      "loss": 0.2466,
+      "step": 944
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.37774738669395447,
+      "learning_rate": 6.0815734428050533e-05,
+      "loss": 0.2342,
+      "step": 945
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.412067711353302,
+      "learning_rate": 6.074076541840977e-05,
+      "loss": 0.3069,
+      "step": 946
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4156152904033661,
+      "learning_rate": 6.066577108527468e-05,
+      "loss": 0.3053,
+      "step": 947
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.403237521648407,
+      "learning_rate": 6.059075160545933e-05,
+      "loss": 0.2943,
+      "step": 948
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5167718529701233,
+      "learning_rate": 6.05157071558371e-05,
+      "loss": 0.4531,
+      "step": 949
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38771021366119385,
+      "learning_rate": 6.0440637913340224e-05,
+      "loss": 0.2165,
+      "step": 950
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.35290205478668213,
+      "learning_rate": 6.03655440549594e-05,
+      "loss": 0.2631,
+      "step": 951
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.37738096714019775,
+      "learning_rate": 6.029042575774334e-05,
+      "loss": 0.2224,
+      "step": 952
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3894968628883362,
+      "learning_rate": 6.0215283198798436e-05,
+      "loss": 0.2039,
+      "step": 953
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.46379074454307556,
+      "learning_rate": 6.0140116555288204e-05,
+      "loss": 0.3499,
+      "step": 954
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4142317473888397,
+      "learning_rate": 6.006492600443301e-05,
+      "loss": 0.2663,
+      "step": 955
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5037583112716675,
+      "learning_rate": 5.998971172350953e-05,
+      "loss": 0.3317,
+      "step": 956
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5221759080886841,
+      "learning_rate": 5.991447388985045e-05,
+      "loss": 0.4407,
+      "step": 957
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.45446208119392395,
+      "learning_rate": 5.9839212680843925e-05,
+      "loss": 0.3386,
+      "step": 958
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.39449796080589294,
+      "learning_rate": 5.9763928273933256e-05,
+      "loss": 0.2174,
+      "step": 959
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4320032298564911,
+      "learning_rate": 5.968862084661643e-05,
+      "loss": 0.3401,
+      "step": 960
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.38568902015686035,
+      "learning_rate": 5.9613290576445704e-05,
+      "loss": 0.2269,
+      "step": 961
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4989192485809326,
+      "learning_rate": 5.953793764102722e-05,
+      "loss": 0.375,
+      "step": 962
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3766809105873108,
+      "learning_rate": 5.946256221802051e-05,
+      "loss": 0.1954,
+      "step": 963
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.40084874629974365,
+      "learning_rate": 5.938716448513818e-05,
+      "loss": 0.3267,
+      "step": 964
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.46326810121536255,
+      "learning_rate": 5.931174462014537e-05,
+      "loss": 0.2894,
+      "step": 965
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.42025160789489746,
+      "learning_rate": 5.923630280085948e-05,
+      "loss": 0.2946,
+      "step": 966
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3362591862678528,
+      "learning_rate": 5.916083920514959e-05,
+      "loss": 0.1692,
+      "step": 967
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3418210744857788,
+      "learning_rate": 5.908535401093618e-05,
+      "loss": 0.1921,
+      "step": 968
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.458325058221817,
+      "learning_rate": 5.900984739619062e-05,
+      "loss": 0.3589,
+      "step": 969
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.34943440556526184,
+      "learning_rate": 5.8934319538934824e-05,
+      "loss": 0.2017,
+      "step": 970
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.42351827025413513,
+      "learning_rate": 5.8858770617240746e-05,
+      "loss": 0.2601,
+      "step": 971
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.585625946521759,
+      "learning_rate": 5.878320080923001e-05,
+      "loss": 0.4344,
+      "step": 972
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5019210577011108,
+      "learning_rate": 5.870761029307352e-05,
+      "loss": 0.3614,
+      "step": 973
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.36957842111587524,
+      "learning_rate": 5.863199924699095e-05,
+      "loss": 0.23,
+      "step": 974
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.39462417364120483,
+      "learning_rate": 5.8556367849250436e-05,
+      "loss": 0.287,
+      "step": 975
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3435553312301636,
+      "learning_rate": 5.848071627816803e-05,
+      "loss": 0.1963,
+      "step": 976
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.41443926095962524,
+      "learning_rate": 5.840504471210742e-05,
+      "loss": 0.3484,
+      "step": 977
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.48773089051246643,
+      "learning_rate": 5.8329353329479366e-05,
+      "loss": 0.3624,
+      "step": 978
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4465321898460388,
+      "learning_rate": 5.82536423087414e-05,
+      "loss": 0.2871,
+      "step": 979
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3528132140636444,
+      "learning_rate": 5.8177911828397336e-05,
+      "loss": 0.169,
+      "step": 980
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.448798269033432,
+      "learning_rate": 5.8102162066996854e-05,
+      "loss": 0.3308,
+      "step": 981
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4471529722213745,
+      "learning_rate": 5.802639320313514e-05,
+      "loss": 0.334,
+      "step": 982
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.41685938835144043,
+      "learning_rate": 5.795060541545236e-05,
+      "loss": 0.2614,
+      "step": 983
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5040465593338013,
+      "learning_rate": 5.787479888263333e-05,
+      "loss": 0.3488,
+      "step": 984
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.42671287059783936,
+      "learning_rate": 5.779897378340705e-05,
+      "loss": 0.2985,
+      "step": 985
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4578186571598053,
+      "learning_rate": 5.77231302965463e-05,
+      "loss": 0.3092,
+      "step": 986
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4186210334300995,
+      "learning_rate": 5.76472686008672e-05,
+      "loss": 0.242,
+      "step": 987
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4466264247894287,
+      "learning_rate": 5.757138887522884e-05,
+      "loss": 0.2823,
+      "step": 988
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4987810552120209,
+      "learning_rate": 5.7495491298532765e-05,
+      "loss": 0.3368,
+      "step": 989
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3911457061767578,
+      "learning_rate": 5.7419576049722646e-05,
+      "loss": 0.2505,
+      "step": 990
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.44448885321617126,
+      "learning_rate": 5.73436433077838e-05,
+      "loss": 0.2498,
+      "step": 991
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.37302330136299133,
+      "learning_rate": 5.7267693251742794e-05,
+      "loss": 0.1792,
+      "step": 992
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39692917466163635,
+      "learning_rate": 5.719172606066703e-05,
+      "loss": 0.3098,
+      "step": 993
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4800136685371399,
+      "learning_rate": 5.7115741913664264e-05,
+      "loss": 0.3715,
+      "step": 994
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.44176486134529114,
+      "learning_rate": 5.703974098988229e-05,
+      "loss": 0.3398,
+      "step": 995
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4073839485645294,
+      "learning_rate": 5.696372346850841e-05,
+      "loss": 0.2772,
+      "step": 996
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.33116286993026733,
+      "learning_rate": 5.68876895287691e-05,
+      "loss": 0.184,
+      "step": 997
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5296633243560791,
+      "learning_rate": 5.68116393499295e-05,
+      "loss": 0.3604,
+      "step": 998
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.46189868450164795,
+      "learning_rate": 5.673557311129306e-05,
+      "loss": 0.3048,
+      "step": 999
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3623585104942322,
+      "learning_rate": 5.6659490992201094e-05,
+      "loss": 0.1948,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.2738782465457916,
+      "eval_runtime": 193.2652,
+      "eval_samples_per_second": 1.764,
+      "eval_steps_per_second": 0.222,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.42749810218811035,
+      "learning_rate": 5.658339317203235e-05,
+      "loss": 0.2543,
+      "step": 1001
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5538439750671387,
+      "learning_rate": 5.650727983020262e-05,
+      "loss": 0.3891,
+      "step": 1002
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3897821605205536,
+      "learning_rate": 5.643115114616425e-05,
+      "loss": 0.2227,
+      "step": 1003
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.469111829996109,
+      "learning_rate": 5.635500729940577e-05,
+      "loss": 0.3361,
+      "step": 1004
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.48245707154273987,
+      "learning_rate": 5.627884846945151e-05,
+      "loss": 0.4045,
+      "step": 1005
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4152395725250244,
+      "learning_rate": 5.620267483586105e-05,
+      "loss": 0.2856,
+      "step": 1006
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4420711398124695,
+      "learning_rate": 5.612648657822892e-05,
+      "loss": 0.3291,
+      "step": 1007
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3620210886001587,
+      "learning_rate": 5.6050283876184116e-05,
+      "loss": 0.2406,
+      "step": 1008
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.42133828997612,
+      "learning_rate": 5.597406690938969e-05,
+      "loss": 0.2757,
+      "step": 1009
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.46080297231674194,
+      "learning_rate": 5.5897835857542317e-05,
+      "loss": 0.291,
+      "step": 1010
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.46735844016075134,
+      "learning_rate": 5.5821590900371886e-05,
+      "loss": 0.2787,
+      "step": 1011
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3542601466178894,
+      "learning_rate": 5.574533221764109e-05,
+      "loss": 0.2515,
+      "step": 1012
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38107067346572876,
+      "learning_rate": 5.566905998914496e-05,
+      "loss": 0.1951,
+      "step": 1013
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4040825366973877,
+      "learning_rate": 5.5592774394710465e-05,
+      "loss": 0.2291,
+      "step": 1014
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4934174716472626,
+      "learning_rate": 5.551647561419611e-05,
+      "loss": 0.207,
+      "step": 1015
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.35920339822769165,
+      "learning_rate": 5.5440163827491454e-05,
+      "loss": 0.2451,
+      "step": 1016
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.41609564423561096,
+      "learning_rate": 5.536383921451673e-05,
+      "loss": 0.2428,
+      "step": 1017
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4347526431083679,
+      "learning_rate": 5.5287501955222444e-05,
+      "loss": 0.3375,
+      "step": 1018
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.41964301466941833,
+      "learning_rate": 5.521115222958888e-05,
+      "loss": 0.2745,
+      "step": 1019
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.39877617359161377,
+      "learning_rate": 5.513479021762572e-05,
+      "loss": 0.242,
+      "step": 1020
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4762844741344452,
+      "learning_rate": 5.505841609937161e-05,
+      "loss": 0.3768,
+      "step": 1021
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5285370349884033,
+      "learning_rate": 5.4982030054893785e-05,
+      "loss": 0.2997,
+      "step": 1022
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.43233630061149597,
+      "learning_rate": 5.490563226428755e-05,
+      "loss": 0.2961,
+      "step": 1023
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4417082965373993,
+      "learning_rate": 5.482922290767589e-05,
+      "loss": 0.2837,
+      "step": 1024
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.32413554191589355,
+      "learning_rate": 5.475280216520913e-05,
+      "loss": 0.1549,
+      "step": 1025
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5058920979499817,
+      "learning_rate": 5.467637021706438e-05,
+      "loss": 0.3349,
+      "step": 1026
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5090081095695496,
+      "learning_rate": 5.4599927243445156e-05,
+      "loss": 0.3377,
+      "step": 1027
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4480152130126953,
+      "learning_rate": 5.452347342458104e-05,
+      "loss": 0.2672,
+      "step": 1028
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.41146597266197205,
+      "learning_rate": 5.4447008940727115e-05,
+      "loss": 0.2598,
+      "step": 1029
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.43932464718818665,
+      "learning_rate": 5.437053397216364e-05,
+      "loss": 0.294,
+      "step": 1030
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.47837650775909424,
+      "learning_rate": 5.429404869919559e-05,
+      "loss": 0.3147,
+      "step": 1031
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5078743100166321,
+      "learning_rate": 5.4217553302152237e-05,
+      "loss": 0.3581,
+      "step": 1032
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4359373450279236,
+      "learning_rate": 5.414104796138673e-05,
+      "loss": 0.2841,
+      "step": 1033
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.42723405361175537,
+      "learning_rate": 5.406453285727564e-05,
+      "loss": 0.2448,
+      "step": 1034
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3603673279285431,
+      "learning_rate": 5.398800817021856e-05,
+      "loss": 0.1881,
+      "step": 1035
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.48210036754608154,
+      "learning_rate": 5.39114740806377e-05,
+      "loss": 0.3464,
+      "step": 1036
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5001813769340515,
+      "learning_rate": 5.383493076897742e-05,
+      "loss": 0.3895,
+      "step": 1037
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.41803812980651855,
+      "learning_rate": 5.3758378415703825e-05,
+      "loss": 0.2698,
+      "step": 1038
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4540984332561493,
+      "learning_rate": 5.3681817201304344e-05,
+      "loss": 0.2462,
+      "step": 1039
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4440363645553589,
+      "learning_rate": 5.360524730628728e-05,
+      "loss": 0.3137,
+      "step": 1040
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3853364884853363,
+      "learning_rate": 5.352866891118142e-05,
+      "loss": 0.2611,
+      "step": 1041
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.38266053795814514,
+      "learning_rate": 5.345208219653561e-05,
+      "loss": 0.2741,
+      "step": 1042
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.350251168012619,
+      "learning_rate": 5.337548734291826e-05,
+      "loss": 0.154,
+      "step": 1043
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3970303535461426,
+      "learning_rate": 5.3298884530917014e-05,
+      "loss": 0.2541,
+      "step": 1044
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5553082227706909,
+      "learning_rate": 5.322227394113825e-05,
+      "loss": 0.4557,
+      "step": 1045
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.40032169222831726,
+      "learning_rate": 5.3145655754206714e-05,
+      "loss": 0.2939,
+      "step": 1046
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.47654539346694946,
+      "learning_rate": 5.306903015076502e-05,
+      "loss": 0.384,
+      "step": 1047
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.41393399238586426,
+      "learning_rate": 5.2992397311473316e-05,
+      "loss": 0.245,
+      "step": 1048
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.37668856978416443,
+      "learning_rate": 5.291575741700877e-05,
+      "loss": 0.2272,
+      "step": 1049
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.435578852891922,
+      "learning_rate": 5.2839110648065214e-05,
+      "loss": 0.3156,
+      "step": 1050
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.37359413504600525,
+      "learning_rate": 5.276245718535269e-05,
+      "loss": 0.2306,
+      "step": 1051
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.43228819966316223,
+      "learning_rate": 5.2685797209596976e-05,
+      "loss": 0.2848,
+      "step": 1052
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.38589340448379517,
+      "learning_rate": 5.260913090153927e-05,
+      "loss": 0.2502,
+      "step": 1053
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.36114946007728577,
+      "learning_rate": 5.2532458441935636e-05,
+      "loss": 0.2203,
+      "step": 1054
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3997515141963959,
+      "learning_rate": 5.24557800115567e-05,
+      "loss": 0.2904,
+      "step": 1055
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.34927910566329956,
+      "learning_rate": 5.2379095791187124e-05,
+      "loss": 0.2091,
+      "step": 1056
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.39250612258911133,
+      "learning_rate": 5.2302405961625225e-05,
+      "loss": 0.2266,
+      "step": 1057
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3459760844707489,
+      "learning_rate": 5.222571070368257e-05,
+      "loss": 0.2378,
+      "step": 1058
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.382813960313797,
+      "learning_rate": 5.214901019818352e-05,
+      "loss": 0.2153,
+      "step": 1059
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.366940975189209,
+      "learning_rate": 5.2072304625964785e-05,
+      "loss": 0.2065,
+      "step": 1060
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.372653603553772,
+      "learning_rate": 5.199559416787503e-05,
+      "loss": 0.2156,
+      "step": 1061
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.38231584429740906,
+      "learning_rate": 5.191887900477443e-05,
+      "loss": 0.2275,
+      "step": 1062
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3773864805698395,
+      "learning_rate": 5.18421593175343e-05,
+      "loss": 0.2138,
+      "step": 1063
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.38089412450790405,
+      "learning_rate": 5.176543528703657e-05,
+      "loss": 0.2378,
+      "step": 1064
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3438931405544281,
+      "learning_rate": 5.168870709417342e-05,
+      "loss": 0.1671,
+      "step": 1065
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.37076058983802795,
+      "learning_rate": 5.161197491984684e-05,
+      "loss": 0.1954,
+      "step": 1066
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3951203227043152,
+      "learning_rate": 5.153523894496826e-05,
+      "loss": 0.1933,
+      "step": 1067
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.328781396150589,
+      "learning_rate": 5.1458499350458e-05,
+      "loss": 0.1492,
+      "step": 1068
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.4342442452907562,
+      "learning_rate": 5.1381756317244945e-05,
+      "loss": 0.2705,
+      "step": 1069
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.40166395902633667,
+      "learning_rate": 5.130501002626609e-05,
+      "loss": 0.2222,
+      "step": 1070
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.33444875478744507,
+      "learning_rate": 5.1228260658466096e-05,
+      "loss": 0.1813,
+      "step": 1071
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3201729953289032,
+      "learning_rate": 5.11515083947969e-05,
+      "loss": 0.1088,
+      "step": 1072
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.41369521617889404,
+      "learning_rate": 5.107475341621726e-05,
+      "loss": 0.2164,
+      "step": 1073
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4451213479042053,
+      "learning_rate": 5.099799590369231e-05,
+      "loss": 0.2174,
+      "step": 1074
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4616295099258423,
+      "learning_rate": 5.092123603819318e-05,
+      "loss": 0.2736,
+      "step": 1075
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4529765844345093,
+      "learning_rate": 5.084447400069655e-05,
+      "loss": 0.158,
+      "step": 1076
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.44757503271102905,
+      "learning_rate": 5.0767709972184244e-05,
+      "loss": 0.2219,
+      "step": 1077
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3903699815273285,
+      "learning_rate": 5.069094413364272e-05,
+      "loss": 0.1682,
+      "step": 1078
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.34121572971343994,
+      "learning_rate": 5.061417666606274e-05,
+      "loss": 0.112,
+      "step": 1079
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3351310193538666,
+      "learning_rate": 5.053740775043892e-05,
+      "loss": 0.1381,
+      "step": 1080
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4937795400619507,
+      "learning_rate": 5.046063756776925e-05,
+      "loss": 0.2261,
+      "step": 1081
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.38824784755706787,
+      "learning_rate": 5.038386629905475e-05,
+      "loss": 0.1613,
+      "step": 1082
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.34187155961990356,
+      "learning_rate": 5.0307094125298956e-05,
+      "loss": 0.1291,
+      "step": 1083
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.503046452999115,
+      "learning_rate": 5.023032122750759e-05,
+      "loss": 0.234,
+      "step": 1084
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.47818058729171753,
+      "learning_rate": 5.015354778668805e-05,
+      "loss": 0.1944,
+      "step": 1085
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5163326859474182,
+      "learning_rate": 5.007677398384901e-05,
+      "loss": 0.2233,
+      "step": 1086
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5666615962982178,
+      "learning_rate": 5e-05,
+      "loss": 0.1881,
+      "step": 1087
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.4332274794578552,
+      "learning_rate": 4.9923226016151e-05,
+      "loss": 0.1402,
+      "step": 1088
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.47112804651260376,
+      "learning_rate": 4.984645221331196e-05,
+      "loss": 0.1798,
+      "step": 1089
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5971799492835999,
+      "learning_rate": 4.976967877249242e-05,
+      "loss": 0.2773,
+      "step": 1090
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.6237249374389648,
+      "learning_rate": 4.9692905874701055e-05,
+      "loss": 0.2213,
+      "step": 1091
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.6739081740379333,
+      "learning_rate": 4.961613370094527e-05,
+      "loss": 0.2675,
+      "step": 1092
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.47461745142936707,
+      "learning_rate": 4.953936243223076e-05,
+      "loss": 0.1648,
+      "step": 1093
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.4593389630317688,
+      "learning_rate": 4.9462592249561095e-05,
+      "loss": 0.1134,
+      "step": 1094
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.42554348707199097,
+      "learning_rate": 4.938582333393727e-05,
+      "loss": 0.1321,
+      "step": 1095
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7614420652389526,
+      "learning_rate": 4.9309055866357295e-05,
+      "loss": 0.1809,
+      "step": 1096
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5542395710945129,
+      "learning_rate": 4.923229002781577e-05,
+      "loss": 0.2042,
+      "step": 1097
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5410667061805725,
+      "learning_rate": 4.915552599930345e-05,
+      "loss": 0.2044,
+      "step": 1098
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.40755757689476013,
+      "learning_rate": 4.907876396180684e-05,
+      "loss": 0.1221,
+      "step": 1099
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.6663229465484619,
+      "learning_rate": 4.900200409630771e-05,
+      "loss": 0.2424,
+      "step": 1100
+    },
+    {
+      "epoch": 1.04,
+      "eval_loss": 0.2852824628353119,
+      "eval_runtime": 197.8769,
+      "eval_samples_per_second": 1.723,
+      "eval_steps_per_second": 0.217,
+      "step": 1100
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5553132891654968,
+      "learning_rate": 4.892524658378276e-05,
+      "loss": 0.2035,
+      "step": 1101
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.45047542452812195,
+      "learning_rate": 4.884849160520311e-05,
+      "loss": 0.1481,
+      "step": 1102
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5559927225112915,
+      "learning_rate": 4.8771739341533916e-05,
+      "loss": 0.1949,
+      "step": 1103
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.39574360847473145,
+      "learning_rate": 4.869498997373393e-05,
+      "loss": 0.1262,
+      "step": 1104
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5600495934486389,
+      "learning_rate": 4.8618243682755074e-05,
+      "loss": 0.1792,
+      "step": 1105
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6430624127388,
+      "learning_rate": 4.8541500649542014e-05,
+      "loss": 0.253,
+      "step": 1106
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.45293447375297546,
+      "learning_rate": 4.8464761055031755e-05,
+      "loss": 0.1642,
+      "step": 1107
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5247792601585388,
+      "learning_rate": 4.838802508015316e-05,
+      "loss": 0.1656,
+      "step": 1108
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5917544364929199,
+      "learning_rate": 4.831129290582659e-05,
+      "loss": 0.2668,
+      "step": 1109
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.39250388741493225,
+      "learning_rate": 4.823456471296344e-05,
+      "loss": 0.1245,
+      "step": 1110
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5028802752494812,
+      "learning_rate": 4.815784068246571e-05,
+      "loss": 0.2214,
+      "step": 1111
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5044224858283997,
+      "learning_rate": 4.8081120995225574e-05,
+      "loss": 0.2355,
+      "step": 1112
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.46498697996139526,
+      "learning_rate": 4.800440583212499e-05,
+      "loss": 0.1653,
+      "step": 1113
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.4435157775878906,
+      "learning_rate": 4.792769537403523e-05,
+      "loss": 0.1469,
+      "step": 1114
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5212072730064392,
+      "learning_rate": 4.785098980181649e-05,
+      "loss": 0.1877,
+      "step": 1115
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.45803627371788025,
+      "learning_rate": 4.777428929631743e-05,
+      "loss": 0.1374,
+      "step": 1116
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.4896143972873688,
+      "learning_rate": 4.7697594038374786e-05,
+      "loss": 0.1914,
+      "step": 1117
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.43538233637809753,
+      "learning_rate": 4.762090420881289e-05,
+      "loss": 0.1671,
+      "step": 1118
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5436176061630249,
+      "learning_rate": 4.754421998844331e-05,
+      "loss": 0.2433,
+      "step": 1119
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.4804631173610687,
+      "learning_rate": 4.746754155806437e-05,
+      "loss": 0.2021,
+      "step": 1120
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5616614818572998,
+      "learning_rate": 4.7390869098460745e-05,
+      "loss": 0.2403,
+      "step": 1121
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.4624713957309723,
+      "learning_rate": 4.731420279040303e-05,
+      "loss": 0.1906,
+      "step": 1122
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5428637266159058,
+      "learning_rate": 4.7237542814647325e-05,
+      "loss": 0.2141,
+      "step": 1123
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4959060251712799,
+      "learning_rate": 4.716088935193479e-05,
+      "loss": 0.1829,
+      "step": 1124
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4145790636539459,
+      "learning_rate": 4.708424258299124e-05,
+      "loss": 0.1668,
+      "step": 1125
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.40106552839279175,
+      "learning_rate": 4.7007602688526695e-05,
+      "loss": 0.1482,
+      "step": 1126
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4693151116371155,
+      "learning_rate": 4.693096984923499e-05,
+      "loss": 0.1872,
+      "step": 1127
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.42648446559906006,
+      "learning_rate": 4.6854344245793305e-05,
+      "loss": 0.1125,
+      "step": 1128
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5151666402816772,
+      "learning_rate": 4.677772605886175e-05,
+      "loss": 0.1956,
+      "step": 1129
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.42513900995254517,
+      "learning_rate": 4.6701115469083e-05,
+      "loss": 0.1363,
+      "step": 1130
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5178140997886658,
+      "learning_rate": 4.662451265708174e-05,
+      "loss": 0.1736,
+      "step": 1131
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.46684348583221436,
+      "learning_rate": 4.65479178034644e-05,
+      "loss": 0.1236,
+      "step": 1132
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5476909279823303,
+      "learning_rate": 4.647133108881858e-05,
+      "loss": 0.1492,
+      "step": 1133
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.44335708022117615,
+      "learning_rate": 4.6394752693712734e-05,
+      "loss": 0.1544,
+      "step": 1134
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5809124708175659,
+      "learning_rate": 4.631818279869566e-05,
+      "loss": 0.2857,
+      "step": 1135
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.523441731929779,
+      "learning_rate": 4.624162158429619e-05,
+      "loss": 0.2217,
+      "step": 1136
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.524860143661499,
+      "learning_rate": 4.616506923102259e-05,
+      "loss": 0.1619,
+      "step": 1137
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5173048973083496,
+      "learning_rate": 4.608852591936231e-05,
+      "loss": 0.1539,
+      "step": 1138
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.36878007650375366,
+      "learning_rate": 4.6011991829781456e-05,
+      "loss": 0.0964,
+      "step": 1139
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.587944507598877,
+      "learning_rate": 4.593546714272438e-05,
+      "loss": 0.183,
+      "step": 1140
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6191943883895874,
+      "learning_rate": 4.585895203861328e-05,
+      "loss": 0.2522,
+      "step": 1141
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.4109521806240082,
+      "learning_rate": 4.578244669784777e-05,
+      "loss": 0.1145,
+      "step": 1142
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5277302265167236,
+      "learning_rate": 4.5705951300804424e-05,
+      "loss": 0.1898,
+      "step": 1143
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5482195019721985,
+      "learning_rate": 4.562946602783636e-05,
+      "loss": 0.1962,
+      "step": 1144
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.45670923590660095,
+      "learning_rate": 4.55529910592729e-05,
+      "loss": 0.1866,
+      "step": 1145
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.6069998145103455,
+      "learning_rate": 4.547652657541897e-05,
+      "loss": 0.272,
+      "step": 1146
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.531523585319519,
+      "learning_rate": 4.540007275655485e-05,
+      "loss": 0.1954,
+      "step": 1147
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.4606556296348572,
+      "learning_rate": 4.532362978293564e-05,
+      "loss": 0.1575,
+      "step": 1148
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.606295645236969,
+      "learning_rate": 4.5247197834790876e-05,
+      "loss": 0.2268,
+      "step": 1149
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5138577818870544,
+      "learning_rate": 4.517077709232411e-05,
+      "loss": 0.2063,
+      "step": 1150
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5193535089492798,
+      "learning_rate": 4.509436773571247e-05,
+      "loss": 0.2221,
+      "step": 1151
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5370643734931946,
+      "learning_rate": 4.501796994510622e-05,
+      "loss": 0.2211,
+      "step": 1152
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.46137669682502747,
+      "learning_rate": 4.49415839006284e-05,
+      "loss": 0.1948,
+      "step": 1153
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.46311068534851074,
+      "learning_rate": 4.4865209782374304e-05,
+      "loss": 0.1268,
+      "step": 1154
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.48577478528022766,
+      "learning_rate": 4.478884777041115e-05,
+      "loss": 0.1804,
+      "step": 1155
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4366888999938965,
+      "learning_rate": 4.471249804477758e-05,
+      "loss": 0.1458,
+      "step": 1156
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.44861680269241333,
+      "learning_rate": 4.463616078548328e-05,
+      "loss": 0.1624,
+      "step": 1157
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4577074944972992,
+      "learning_rate": 4.4559836172508565e-05,
+      "loss": 0.1551,
+      "step": 1158
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5058170557022095,
+      "learning_rate": 4.448352438580391e-05,
+      "loss": 0.1872,
+      "step": 1159
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.37819817662239075,
+      "learning_rate": 4.4407225605289547e-05,
+      "loss": 0.1293,
+      "step": 1160
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5454598665237427,
+      "learning_rate": 4.433094001085505e-05,
+      "loss": 0.2349,
+      "step": 1161
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.49581095576286316,
+      "learning_rate": 4.4254667782358924e-05,
+      "loss": 0.1665,
+      "step": 1162
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.49631190299987793,
+      "learning_rate": 4.417840909962813e-05,
+      "loss": 0.145,
+      "step": 1163
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4741055369377136,
+      "learning_rate": 4.410216414245771e-05,
+      "loss": 0.1687,
+      "step": 1164
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5791394114494324,
+      "learning_rate": 4.402593309061034e-05,
+      "loss": 0.2523,
+      "step": 1165
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4387664198875427,
+      "learning_rate": 4.394971612381591e-05,
+      "loss": 0.1418,
+      "step": 1166
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5703561902046204,
+      "learning_rate": 4.38735134217711e-05,
+      "loss": 0.1734,
+      "step": 1167
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5360005497932434,
+      "learning_rate": 4.379732516413897e-05,
+      "loss": 0.2169,
+      "step": 1168
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5344962477684021,
+      "learning_rate": 4.37211515305485e-05,
+      "loss": 0.2319,
+      "step": 1169
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5229570865631104,
+      "learning_rate": 4.364499270059424e-05,
+      "loss": 0.2057,
+      "step": 1170
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6585649847984314,
+      "learning_rate": 4.3568848853835774e-05,
+      "loss": 0.2493,
+      "step": 1171
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6584741473197937,
+      "learning_rate": 4.34927201697974e-05,
+      "loss": 0.2311,
+      "step": 1172
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.4309971332550049,
+      "learning_rate": 4.341660682796766e-05,
+      "loss": 0.1479,
+      "step": 1173
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.4658047556877136,
+      "learning_rate": 4.334050900779893e-05,
+      "loss": 0.1484,
+      "step": 1174
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.45378291606903076,
+      "learning_rate": 4.3264426888706965e-05,
+      "loss": 0.1466,
+      "step": 1175
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.542150616645813,
+      "learning_rate": 4.3188360650070525e-05,
+      "loss": 0.2085,
+      "step": 1176
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.4845476448535919,
+      "learning_rate": 4.311231047123092e-05,
+      "loss": 0.1974,
+      "step": 1177
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.461410254240036,
+      "learning_rate": 4.303627653149159e-05,
+      "loss": 0.1411,
+      "step": 1178
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.49094563722610474,
+      "learning_rate": 4.2960259010117724e-05,
+      "loss": 0.1797,
+      "step": 1179
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5522293448448181,
+      "learning_rate": 4.288425808633575e-05,
+      "loss": 0.1881,
+      "step": 1180
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.548546552658081,
+      "learning_rate": 4.2808273939333e-05,
+      "loss": 0.1692,
+      "step": 1181
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5489959716796875,
+      "learning_rate": 4.2732306748257224e-05,
+      "loss": 0.1785,
+      "step": 1182
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5799089670181274,
+      "learning_rate": 4.265635669221622e-05,
+      "loss": 0.2268,
+      "step": 1183
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5645148754119873,
+      "learning_rate": 4.258042395027738e-05,
+      "loss": 0.2222,
+      "step": 1184
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.40521708130836487,
+      "learning_rate": 4.250450870146725e-05,
+      "loss": 0.125,
+      "step": 1185
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.45114371180534363,
+      "learning_rate": 4.2428611124771184e-05,
+      "loss": 0.145,
+      "step": 1186
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5046716332435608,
+      "learning_rate": 4.2352731399132804e-05,
+      "loss": 0.1855,
+      "step": 1187
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5100313425064087,
+      "learning_rate": 4.2276869703453724e-05,
+      "loss": 0.2132,
+      "step": 1188
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.46480703353881836,
+      "learning_rate": 4.2201026216592975e-05,
+      "loss": 0.1714,
+      "step": 1189
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.4161708950996399,
+      "learning_rate": 4.21252011173667e-05,
+      "loss": 0.1363,
+      "step": 1190
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5647803544998169,
+      "learning_rate": 4.2049394584547664e-05,
+      "loss": 0.1779,
+      "step": 1191
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.4570713937282562,
+      "learning_rate": 4.1973606796864884e-05,
+      "loss": 0.1488,
+      "step": 1192
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5394353866577148,
+      "learning_rate": 4.189783793300316e-05,
+      "loss": 0.2258,
+      "step": 1193
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5679755806922913,
+      "learning_rate": 4.182208817160269e-05,
+      "loss": 0.19,
+      "step": 1194
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5621250867843628,
+      "learning_rate": 4.1746357691258614e-05,
+      "loss": 0.1762,
+      "step": 1195
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.585514485836029,
+      "learning_rate": 4.167064667052065e-05,
+      "loss": 0.1977,
+      "step": 1196
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.44670602679252625,
+      "learning_rate": 4.15949552878926e-05,
+      "loss": 0.1536,
+      "step": 1197
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.48966893553733826,
+      "learning_rate": 4.151928372183198e-05,
+      "loss": 0.1725,
+      "step": 1198
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.4429165720939636,
+      "learning_rate": 4.144363215074958e-05,
+      "loss": 0.1678,
+      "step": 1199
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.3323366641998291,
+      "learning_rate": 4.136800075300906e-05,
+      "loss": 0.1003,
+      "step": 1200
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 0.2810652554035187,
+      "eval_runtime": 196.3019,
+      "eval_samples_per_second": 1.737,
+      "eval_steps_per_second": 0.219,
+      "step": 1200
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5352948904037476,
+      "learning_rate": 4.12923897069265e-05,
+      "loss": 0.1227,
+      "step": 1201
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.514959454536438,
+      "learning_rate": 4.1216799190770006e-05,
+      "loss": 0.1891,
+      "step": 1202
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5169869661331177,
+      "learning_rate": 4.1141229382759286e-05,
+      "loss": 0.2232,
+      "step": 1203
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5219796299934387,
+      "learning_rate": 4.1065680461065194e-05,
+      "loss": 0.2065,
+      "step": 1204
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.46082746982574463,
+      "learning_rate": 4.099015260380939e-05,
+      "loss": 0.1434,
+      "step": 1205
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5533954501152039,
+      "learning_rate": 4.0914645989063845e-05,
+      "loss": 0.2617,
+      "step": 1206
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.47323837876319885,
+      "learning_rate": 4.0839160794850436e-05,
+      "loss": 0.1858,
+      "step": 1207
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.4898775517940521,
+      "learning_rate": 4.076369719914055e-05,
+      "loss": 0.1727,
+      "step": 1208
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5496148467063904,
+      "learning_rate": 4.068825537985465e-05,
+      "loss": 0.2369,
+      "step": 1209
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5217909216880798,
+      "learning_rate": 4.061283551486185e-05,
+      "loss": 0.2296,
+      "step": 1210
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5044183731079102,
+      "learning_rate": 4.0537437781979506e-05,
+      "loss": 0.1793,
+      "step": 1211
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5007127523422241,
+      "learning_rate": 4.0462062358972796e-05,
+      "loss": 0.2047,
+      "step": 1212
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.37619128823280334,
+      "learning_rate": 4.038670942355431e-05,
+      "loss": 0.1119,
+      "step": 1213
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.48169299960136414,
+      "learning_rate": 4.03113791533836e-05,
+      "loss": 0.1671,
+      "step": 1214
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.46656301617622375,
+      "learning_rate": 4.023607172606676e-05,
+      "loss": 0.1491,
+      "step": 1215
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5641793012619019,
+      "learning_rate": 4.016078731915608e-05,
+      "loss": 0.2129,
+      "step": 1216
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.491321861743927,
+      "learning_rate": 4.008552611014955e-05,
+      "loss": 0.1894,
+      "step": 1217
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.4718596339225769,
+      "learning_rate": 4.001028827649046e-05,
+      "loss": 0.1596,
+      "step": 1218
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.46768590807914734,
+      "learning_rate": 3.993507399556699e-05,
+      "loss": 0.163,
+      "step": 1219
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.6066141724586487,
+      "learning_rate": 3.9859883444711794e-05,
+      "loss": 0.2633,
+      "step": 1220
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5247323513031006,
+      "learning_rate": 3.978471680120157e-05,
+      "loss": 0.2386,
+      "step": 1221
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.43791836500167847,
+      "learning_rate": 3.970957424225666e-05,
+      "loss": 0.176,
+      "step": 1222
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.35836443305015564,
+      "learning_rate": 3.9634455945040614e-05,
+      "loss": 0.1244,
+      "step": 1223
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5778273344039917,
+      "learning_rate": 3.955936208665979e-05,
+      "loss": 0.1913,
+      "step": 1224
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.6201069355010986,
+      "learning_rate": 3.9484292844162904e-05,
+      "loss": 0.2293,
+      "step": 1225
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.4196273386478424,
+      "learning_rate": 3.940924839454066e-05,
+      "loss": 0.156,
+      "step": 1226
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.4268430173397064,
+      "learning_rate": 3.933422891472532e-05,
+      "loss": 0.1658,
+      "step": 1227
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5689937472343445,
+      "learning_rate": 3.925923458159023e-05,
+      "loss": 0.2288,
+      "step": 1228
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5220311284065247,
+      "learning_rate": 3.918426557194947e-05,
+      "loss": 0.2111,
+      "step": 1229
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5138110518455505,
+      "learning_rate": 3.9109322062557424e-05,
+      "loss": 0.1955,
+      "step": 1230
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5784496665000916,
+      "learning_rate": 3.903440423010834e-05,
+      "loss": 0.2645,
+      "step": 1231
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5086357593536377,
+      "learning_rate": 3.895951225123595e-05,
+      "loss": 0.181,
+      "step": 1232
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.49407991766929626,
+      "learning_rate": 3.888464630251298e-05,
+      "loss": 0.1975,
+      "step": 1233
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5430823564529419,
+      "learning_rate": 3.880980656045087e-05,
+      "loss": 0.2133,
+      "step": 1234
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5606887936592102,
+      "learning_rate": 3.873499320149918e-05,
+      "loss": 0.2534,
+      "step": 1235
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5339627265930176,
+      "learning_rate": 3.866020640204533e-05,
+      "loss": 0.2436,
+      "step": 1236
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.43829765915870667,
+      "learning_rate": 3.858544633841409e-05,
+      "loss": 0.163,
+      "step": 1237
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.4846597909927368,
+      "learning_rate": 3.851071318686721e-05,
+      "loss": 0.1758,
+      "step": 1238
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5583491921424866,
+      "learning_rate": 3.843600712360298e-05,
+      "loss": 0.1939,
+      "step": 1239
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.4945659041404724,
+      "learning_rate": 3.8361328324755825e-05,
+      "loss": 0.1857,
+      "step": 1240
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.4546511173248291,
+      "learning_rate": 3.82866769663959e-05,
+      "loss": 0.1426,
+      "step": 1241
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.4968038499355316,
+      "learning_rate": 3.821205322452863e-05,
+      "loss": 0.186,
+      "step": 1242
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6010574102401733,
+      "learning_rate": 3.813745727509439e-05,
+      "loss": 0.2333,
+      "step": 1243
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5425085425376892,
+      "learning_rate": 3.8062889293967976e-05,
+      "loss": 0.182,
+      "step": 1244
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.49743250012397766,
+      "learning_rate": 3.798834945695826e-05,
+      "loss": 0.2018,
+      "step": 1245
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.46727320551872253,
+      "learning_rate": 3.791383793980776e-05,
+      "loss": 0.1588,
+      "step": 1246
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5810048580169678,
+      "learning_rate": 3.7839354918192215e-05,
+      "loss": 0.2053,
+      "step": 1247
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.4939357340335846,
+      "learning_rate": 3.7764900567720196e-05,
+      "loss": 0.1865,
+      "step": 1248
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5220310688018799,
+      "learning_rate": 3.769047506393267e-05,
+      "loss": 0.2499,
+      "step": 1249
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.4659373462200165,
+      "learning_rate": 3.761607858230257e-05,
+      "loss": 0.1517,
+      "step": 1250
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5804378986358643,
+      "learning_rate": 3.7541711298234436e-05,
+      "loss": 0.2214,
+      "step": 1251
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5237458348274231,
+      "learning_rate": 3.746737338706397e-05,
+      "loss": 0.1459,
+      "step": 1252
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.4977158010005951,
+      "learning_rate": 3.7393065024057596e-05,
+      "loss": 0.181,
+      "step": 1253
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.4609799385070801,
+      "learning_rate": 3.7318786384412074e-05,
+      "loss": 0.1696,
+      "step": 1254
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5353249907493591,
+      "learning_rate": 3.724453764325411e-05,
+      "loss": 0.2099,
+      "step": 1255
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5042243003845215,
+      "learning_rate": 3.71703189756399e-05,
+      "loss": 0.1847,
+      "step": 1256
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5251048803329468,
+      "learning_rate": 3.709613055655474e-05,
+      "loss": 0.1725,
+      "step": 1257
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6426075100898743,
+      "learning_rate": 3.70219725609126e-05,
+      "loss": 0.2448,
+      "step": 1258
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.4450177848339081,
+      "learning_rate": 3.694784516355573e-05,
+      "loss": 0.1557,
+      "step": 1259
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.40712741017341614,
+      "learning_rate": 3.6873748539254246e-05,
+      "loss": 0.1226,
+      "step": 1260
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.39405110478401184,
+      "learning_rate": 3.679968286270571e-05,
+      "loss": 0.122,
+      "step": 1261
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.4739035367965698,
+      "learning_rate": 3.6725648308534696e-05,
+      "loss": 0.1891,
+      "step": 1262
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.50302654504776,
+      "learning_rate": 3.6651645051292415e-05,
+      "loss": 0.1609,
+      "step": 1263
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.4972444474697113,
+      "learning_rate": 3.65776732654563e-05,
+      "loss": 0.1909,
+      "step": 1264
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.7175520658493042,
+      "learning_rate": 3.650373312542956e-05,
+      "loss": 0.344,
+      "step": 1265
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.528699517250061,
+      "learning_rate": 3.6429824805540815e-05,
+      "loss": 0.1731,
+      "step": 1266
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.4887191355228424,
+      "learning_rate": 3.6355948480043644e-05,
+      "loss": 0.1553,
+      "step": 1267
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.4959406554698944,
+      "learning_rate": 3.628210432311621e-05,
+      "loss": 0.1909,
+      "step": 1268
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5369499325752258,
+      "learning_rate": 3.620829250886083e-05,
+      "loss": 0.2713,
+      "step": 1269
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5405216813087463,
+      "learning_rate": 3.6134513211303556e-05,
+      "loss": 0.2235,
+      "step": 1270
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5250999331474304,
+      "learning_rate": 3.606076660439378e-05,
+      "loss": 0.1755,
+      "step": 1271
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.4903579354286194,
+      "learning_rate": 3.598705286200382e-05,
+      "loss": 0.2088,
+      "step": 1272
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5387226939201355,
+      "learning_rate": 3.591337215792852e-05,
+      "loss": 0.2429,
+      "step": 1273
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.6268158555030823,
+      "learning_rate": 3.58397246658848e-05,
+      "loss": 0.1536,
+      "step": 1274
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5656338930130005,
+      "learning_rate": 3.576611055951131e-05,
+      "loss": 0.2089,
+      "step": 1275
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.6027837991714478,
+      "learning_rate": 3.5692530012367955e-05,
+      "loss": 0.1992,
+      "step": 1276
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5068827867507935,
+      "learning_rate": 3.561898319793555e-05,
+      "loss": 0.1984,
+      "step": 1277
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5017920136451721,
+      "learning_rate": 3.554547028961537e-05,
+      "loss": 0.2032,
+      "step": 1278
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5080841779708862,
+      "learning_rate": 3.547199146072873e-05,
+      "loss": 0.1554,
+      "step": 1279
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.6594606637954712,
+      "learning_rate": 3.5398546884516604e-05,
+      "loss": 0.2678,
+      "step": 1280
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5818920731544495,
+      "learning_rate": 3.5325136734139214e-05,
+      "loss": 0.203,
+      "step": 1281
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.49282583594322205,
+      "learning_rate": 3.5251761182675625e-05,
+      "loss": 0.1704,
+      "step": 1282
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.4746369421482086,
+      "learning_rate": 3.5178420403123305e-05,
+      "loss": 0.1302,
+      "step": 1283
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.4937484562397003,
+      "learning_rate": 3.510511456839777e-05,
+      "loss": 0.1384,
+      "step": 1284
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.6375249624252319,
+      "learning_rate": 3.5031843851332104e-05,
+      "loss": 0.212,
+      "step": 1285
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5928176045417786,
+      "learning_rate": 3.495860842467664e-05,
+      "loss": 0.2437,
+      "step": 1286
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5119690299034119,
+      "learning_rate": 3.488540846109849e-05,
+      "loss": 0.1577,
+      "step": 1287
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.48352521657943726,
+      "learning_rate": 3.481224413318114e-05,
+      "loss": 0.1488,
+      "step": 1288
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5847229361534119,
+      "learning_rate": 3.473911561342408e-05,
+      "loss": 0.1906,
+      "step": 1289
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5354794859886169,
+      "learning_rate": 3.4666023074242356e-05,
+      "loss": 0.2343,
+      "step": 1290
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.38159751892089844,
+      "learning_rate": 3.459296668796619e-05,
+      "loss": 0.1245,
+      "step": 1291
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5038556456565857,
+      "learning_rate": 3.4519946626840573e-05,
+      "loss": 0.1893,
+      "step": 1292
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.4917728900909424,
+      "learning_rate": 3.444696306302485e-05,
+      "loss": 0.182,
+      "step": 1293
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.6153894066810608,
+      "learning_rate": 3.4374016168592296e-05,
+      "loss": 0.1918,
+      "step": 1294
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5354846715927124,
+      "learning_rate": 3.4301106115529764e-05,
+      "loss": 0.2024,
+      "step": 1295
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5442001819610596,
+      "learning_rate": 3.422823307573722e-05,
+      "loss": 0.217,
+      "step": 1296
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.45692315697669983,
+      "learning_rate": 3.41553972210274e-05,
+      "loss": 0.1661,
+      "step": 1297
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.4843423664569855,
+      "learning_rate": 3.40825987231253e-05,
+      "loss": 0.1994,
+      "step": 1298
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.3701391816139221,
+      "learning_rate": 3.4009837753667916e-05,
+      "loss": 0.1158,
+      "step": 1299
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5183972120285034,
+      "learning_rate": 3.393711448420372e-05,
+      "loss": 0.1485,
+      "step": 1300
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.27870649099349976,
+      "eval_runtime": 195.3057,
+      "eval_samples_per_second": 1.746,
+      "eval_steps_per_second": 0.22,
+      "step": 1300
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.6307404041290283,
+      "learning_rate": 3.38644290861923e-05,
+      "loss": 0.2078,
+      "step": 1301
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.559257984161377,
+      "learning_rate": 3.379178173100396e-05,
+      "loss": 0.2226,
+      "step": 1302
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.648064911365509,
+      "learning_rate": 3.3719172589919326e-05,
+      "loss": 0.2573,
+      "step": 1303
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5507780909538269,
+      "learning_rate": 3.364660183412892e-05,
+      "loss": 0.1442,
+      "step": 1304
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5900538563728333,
+      "learning_rate": 3.357406963473274e-05,
+      "loss": 0.1521,
+      "step": 1305
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.4049758017063141,
+      "learning_rate": 3.3501576162739904e-05,
+      "loss": 0.1071,
+      "step": 1306
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5189416408538818,
+      "learning_rate": 3.3429121589068215e-05,
+      "loss": 0.1558,
+      "step": 1307
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6036385893821716,
+      "learning_rate": 3.335670608454376e-05,
+      "loss": 0.1803,
+      "step": 1308
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5526834726333618,
+      "learning_rate": 3.3284329819900525e-05,
+      "loss": 0.1696,
+      "step": 1309
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.4837692975997925,
+      "learning_rate": 3.321199296577998e-05,
+      "loss": 0.1457,
+      "step": 1310
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.622710108757019,
+      "learning_rate": 3.3139695692730646e-05,
+      "loss": 0.1977,
+      "step": 1311
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.40247225761413574,
+      "learning_rate": 3.3067438171207766e-05,
+      "loss": 0.0937,
+      "step": 1312
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5706564784049988,
+      "learning_rate": 3.2995220571572846e-05,
+      "loss": 0.2201,
+      "step": 1313
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.456645667552948,
+      "learning_rate": 3.292304306409325e-05,
+      "loss": 0.151,
+      "step": 1314
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4740467965602875,
+      "learning_rate": 3.285090581894185e-05,
+      "loss": 0.1717,
+      "step": 1315
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4996465742588043,
+      "learning_rate": 3.2778809006196564e-05,
+      "loss": 0.2013,
+      "step": 1316
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4037359058856964,
+      "learning_rate": 3.270675279583998e-05,
+      "loss": 0.1088,
+      "step": 1317
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4713441729545593,
+      "learning_rate": 3.263473735775899e-05,
+      "loss": 0.1594,
+      "step": 1318
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5820550322532654,
+      "learning_rate": 3.2562762861744326e-05,
+      "loss": 0.2338,
+      "step": 1319
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5265978574752808,
+      "learning_rate": 3.249082947749019e-05,
+      "loss": 0.1729,
+      "step": 1320
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.44064760208129883,
+      "learning_rate": 3.2418937374593894e-05,
+      "loss": 0.1387,
+      "step": 1321
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5152722001075745,
+      "learning_rate": 3.234708672255538e-05,
+      "loss": 0.2262,
+      "step": 1322
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4608455300331116,
+      "learning_rate": 3.227527769077687e-05,
+      "loss": 0.1213,
+      "step": 1323
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.48065075278282166,
+      "learning_rate": 3.220351044856247e-05,
+      "loss": 0.179,
+      "step": 1324
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.36709660291671753,
+      "learning_rate": 3.2131785165117745e-05,
+      "loss": 0.0911,
+      "step": 1325
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5184856653213501,
+      "learning_rate": 3.206010200954935e-05,
+      "loss": 0.1701,
+      "step": 1326
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5596900582313538,
+      "learning_rate": 3.198846115086459e-05,
+      "loss": 0.2092,
+      "step": 1327
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.596989631652832,
+      "learning_rate": 3.191686275797107e-05,
+      "loss": 0.2659,
+      "step": 1328
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.4322189390659332,
+      "learning_rate": 3.1845306999676275e-05,
+      "loss": 0.1297,
+      "step": 1329
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.4216068983078003,
+      "learning_rate": 3.177379404468715e-05,
+      "loss": 0.1202,
+      "step": 1330
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.45087286829948425,
+      "learning_rate": 3.1702324061609736e-05,
+      "loss": 0.1647,
+      "step": 1331
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5325843095779419,
+      "learning_rate": 3.163089721894876e-05,
+      "loss": 0.1803,
+      "step": 1332
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.43478769063949585,
+      "learning_rate": 3.1559513685107235e-05,
+      "loss": 0.1453,
+      "step": 1333
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.45973649621009827,
+      "learning_rate": 3.148817362838606e-05,
+      "loss": 0.1669,
+      "step": 1334
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.4879058301448822,
+      "learning_rate": 3.141687721698363e-05,
+      "loss": 0.1421,
+      "step": 1335
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.4269010126590729,
+      "learning_rate": 3.134562461899545e-05,
+      "loss": 0.1459,
+      "step": 1336
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.4797835052013397,
+      "learning_rate": 3.127441600241369e-05,
+      "loss": 0.1417,
+      "step": 1337
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.527397871017456,
+      "learning_rate": 3.120325153512687e-05,
+      "loss": 0.2442,
+      "step": 1338
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.4763750731945038,
+      "learning_rate": 3.11321313849194e-05,
+      "loss": 0.194,
+      "step": 1339
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.6854526996612549,
+      "learning_rate": 3.10610557194712e-05,
+      "loss": 0.2626,
+      "step": 1340
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5128159523010254,
+      "learning_rate": 3.0990024706357314e-05,
+      "loss": 0.1272,
+      "step": 1341
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.526768147945404,
+      "learning_rate": 3.091903851304751e-05,
+      "loss": 0.1792,
+      "step": 1342
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5984563827514648,
+      "learning_rate": 3.084809730690587e-05,
+      "loss": 0.2443,
+      "step": 1343
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5484467148780823,
+      "learning_rate": 3.077720125519042e-05,
+      "loss": 0.2353,
+      "step": 1344
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5329880118370056,
+      "learning_rate": 3.070635052505273e-05,
+      "loss": 0.2122,
+      "step": 1345
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.42684468626976013,
+      "learning_rate": 3.063554528353752e-05,
+      "loss": 0.1462,
+      "step": 1346
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.580346405506134,
+      "learning_rate": 3.056478569758225e-05,
+      "loss": 0.1838,
+      "step": 1347
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5334434509277344,
+      "learning_rate": 3.0494071934016737e-05,
+      "loss": 0.1542,
+      "step": 1348
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.48692795634269714,
+      "learning_rate": 3.0423404159562774e-05,
+      "loss": 0.1617,
+      "step": 1349
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.554698646068573,
+      "learning_rate": 3.03527825408337e-05,
+      "loss": 0.1911,
+      "step": 1350
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5713726282119751,
+      "learning_rate": 3.0282207244334082e-05,
+      "loss": 0.1995,
+      "step": 1351
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5384231209754944,
+      "learning_rate": 3.0211678436459212e-05,
+      "loss": 0.1845,
+      "step": 1352
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.42969319224357605,
+      "learning_rate": 3.014119628349482e-05,
+      "loss": 0.141,
+      "step": 1353
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5887604355812073,
+      "learning_rate": 3.0070760951616618e-05,
+      "loss": 0.1834,
+      "step": 1354
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5973981022834778,
+      "learning_rate": 3.0000372606889937e-05,
+      "loss": 0.2192,
+      "step": 1355
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.550570011138916,
+      "learning_rate": 2.993003141526933e-05,
+      "loss": 0.2,
+      "step": 1356
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6673139929771423,
+      "learning_rate": 2.9859737542598155e-05,
+      "loss": 0.1247,
+      "step": 1357
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.4876181483268738,
+      "learning_rate": 2.978949115460824e-05,
+      "loss": 0.1837,
+      "step": 1358
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5515248775482178,
+      "learning_rate": 2.971929241691942e-05,
+      "loss": 0.1709,
+      "step": 1359
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.4875487983226776,
+      "learning_rate": 2.9649141495039223e-05,
+      "loss": 0.1625,
+      "step": 1360
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6222435832023621,
+      "learning_rate": 2.957903855436241e-05,
+      "loss": 0.2247,
+      "step": 1361
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5669927597045898,
+      "learning_rate": 2.950898376017064e-05,
+      "loss": 0.2262,
+      "step": 1362
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5209169387817383,
+      "learning_rate": 2.9438977277632017e-05,
+      "loss": 0.2243,
+      "step": 1363
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5239472389221191,
+      "learning_rate": 2.9369019271800823e-05,
+      "loss": 0.2134,
+      "step": 1364
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5234110355377197,
+      "learning_rate": 2.9299109907616955e-05,
+      "loss": 0.1805,
+      "step": 1365
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.427932471036911,
+      "learning_rate": 2.9229249349905684e-05,
+      "loss": 0.1247,
+      "step": 1366
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5718261003494263,
+      "learning_rate": 2.9159437763377184e-05,
+      "loss": 0.1913,
+      "step": 1367
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5143815279006958,
+      "learning_rate": 2.908967531262618e-05,
+      "loss": 0.2012,
+      "step": 1368
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5429434180259705,
+      "learning_rate": 2.9019962162131565e-05,
+      "loss": 0.1743,
+      "step": 1369
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5124776363372803,
+      "learning_rate": 2.895029847625595e-05,
+      "loss": 0.1788,
+      "step": 1370
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4927649199962616,
+      "learning_rate": 2.8880684419245384e-05,
+      "loss": 0.1649,
+      "step": 1371
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4962637722492218,
+      "learning_rate": 2.8811120155228844e-05,
+      "loss": 0.171,
+      "step": 1372
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.47827842831611633,
+      "learning_rate": 2.874160584821798e-05,
+      "loss": 0.1563,
+      "step": 1373
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5000060796737671,
+      "learning_rate": 2.8672141662106578e-05,
+      "loss": 0.1495,
+      "step": 1374
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5428785681724548,
+      "learning_rate": 2.8602727760670332e-05,
+      "loss": 0.214,
+      "step": 1375
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5262407660484314,
+      "learning_rate": 2.853336430756631e-05,
+      "loss": 0.1882,
+      "step": 1376
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.49476855993270874,
+      "learning_rate": 2.846405146633269e-05,
+      "loss": 0.175,
+      "step": 1377
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5650327801704407,
+      "learning_rate": 2.8394789400388328e-05,
+      "loss": 0.1952,
+      "step": 1378
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5197858810424805,
+      "learning_rate": 2.8325578273032294e-05,
+      "loss": 0.2213,
+      "step": 1379
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5687361359596252,
+      "learning_rate": 2.8256418247443662e-05,
+      "loss": 0.1939,
+      "step": 1380
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.49449479579925537,
+      "learning_rate": 2.8187309486680925e-05,
+      "loss": 0.1599,
+      "step": 1381
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.4630528390407562,
+      "learning_rate": 2.811825215368179e-05,
+      "loss": 0.1724,
+      "step": 1382
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5180186629295349,
+      "learning_rate": 2.8049246411262643e-05,
+      "loss": 0.1633,
+      "step": 1383
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5633207559585571,
+      "learning_rate": 2.798029242211828e-05,
+      "loss": 0.2128,
+      "step": 1384
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5610430240631104,
+      "learning_rate": 2.7911390348821508e-05,
+      "loss": 0.2009,
+      "step": 1385
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5820095539093018,
+      "learning_rate": 2.7842540353822632e-05,
+      "loss": 0.1694,
+      "step": 1386
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.6341986656188965,
+      "learning_rate": 2.777374259944929e-05,
+      "loss": 0.236,
+      "step": 1387
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.529824435710907,
+      "learning_rate": 2.770499724790584e-05,
+      "loss": 0.1811,
+      "step": 1388
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5165717005729675,
+      "learning_rate": 2.7636304461273187e-05,
+      "loss": 0.1697,
+      "step": 1389
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.48876580595970154,
+      "learning_rate": 2.7567664401508225e-05,
+      "loss": 0.1882,
+      "step": 1390
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5847304463386536,
+      "learning_rate": 2.7499077230443603e-05,
+      "loss": 0.1932,
+      "step": 1391
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5882861018180847,
+      "learning_rate": 2.7430543109787222e-05,
+      "loss": 0.1799,
+      "step": 1392
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5151212215423584,
+      "learning_rate": 2.7362062201121918e-05,
+      "loss": 0.176,
+      "step": 1393
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5962074995040894,
+      "learning_rate": 2.729363466590511e-05,
+      "loss": 0.1383,
+      "step": 1394
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5134526491165161,
+      "learning_rate": 2.72252606654683e-05,
+      "loss": 0.1742,
+      "step": 1395
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5640712976455688,
+      "learning_rate": 2.7156940361016864e-05,
+      "loss": 0.1678,
+      "step": 1396
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6279372572898865,
+      "learning_rate": 2.708867391362948e-05,
+      "loss": 0.2074,
+      "step": 1397
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5061882734298706,
+      "learning_rate": 2.702046148425795e-05,
+      "loss": 0.151,
+      "step": 1398
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5750608444213867,
+      "learning_rate": 2.6952303233726627e-05,
+      "loss": 0.1993,
+      "step": 1399
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5341944098472595,
+      "learning_rate": 2.688419932273219e-05,
+      "loss": 0.201,
+      "step": 1400
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 0.28342846035957336,
+      "eval_runtime": 197.3534,
+      "eval_samples_per_second": 1.728,
+      "eval_steps_per_second": 0.218,
+      "step": 1400
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.652698278427124,
+      "learning_rate": 2.6816149911843148e-05,
+      "loss": 0.2356,
+      "step": 1401
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5961298942565918,
+      "learning_rate": 2.6748155161499567e-05,
+      "loss": 0.2019,
+      "step": 1402
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.4567265808582306,
+      "learning_rate": 2.6680215232012628e-05,
+      "loss": 0.1495,
+      "step": 1403
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.49438706040382385,
+      "learning_rate": 2.6612330283564223e-05,
+      "loss": 0.1902,
+      "step": 1404
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5643045902252197,
+      "learning_rate": 2.6544500476206673e-05,
+      "loss": 0.2404,
+      "step": 1405
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5158506035804749,
+      "learning_rate": 2.6476725969862227e-05,
+      "loss": 0.1578,
+      "step": 1406
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5846490263938904,
+      "learning_rate": 2.640900692432282e-05,
+      "loss": 0.189,
+      "step": 1407
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.45488500595092773,
+      "learning_rate": 2.634134349924956e-05,
+      "loss": 0.137,
+      "step": 1408
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5066903829574585,
+      "learning_rate": 2.6273735854172488e-05,
+      "loss": 0.1573,
+      "step": 1409
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.48721280694007874,
+      "learning_rate": 2.6206184148490066e-05,
+      "loss": 0.161,
+      "step": 1410
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.549283504486084,
+      "learning_rate": 2.6138688541468903e-05,
+      "loss": 0.1421,
+      "step": 1411
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6236990094184875,
+      "learning_rate": 2.6071249192243363e-05,
+      "loss": 0.2028,
+      "step": 1412
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.7099291682243347,
+      "learning_rate": 2.6003866259815123e-05,
+      "loss": 0.135,
+      "step": 1413
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5931044220924377,
+      "learning_rate": 2.5936539903052892e-05,
+      "loss": 0.2079,
+      "step": 1414
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5032140016555786,
+      "learning_rate": 2.5869270280691944e-05,
+      "loss": 0.1759,
+      "step": 1415
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5954914689064026,
+      "learning_rate": 2.580205755133384e-05,
+      "loss": 0.2682,
+      "step": 1416
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.36140236258506775,
+      "learning_rate": 2.573490187344596e-05,
+      "loss": 0.0813,
+      "step": 1417
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.7367720603942871,
+      "learning_rate": 2.5667803405361213e-05,
+      "loss": 0.2205,
+      "step": 1418
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5584636330604553,
+      "learning_rate": 2.5600762305277583e-05,
+      "loss": 0.1973,
+      "step": 1419
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6195295453071594,
+      "learning_rate": 2.5533778731257824e-05,
+      "loss": 0.1877,
+      "step": 1420
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.62422776222229,
+      "learning_rate": 2.5466852841229088e-05,
+      "loss": 0.2248,
+      "step": 1421
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6158571243286133,
+      "learning_rate": 2.5399984792982457e-05,
+      "loss": 0.1992,
+      "step": 1422
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.546989381313324,
+      "learning_rate": 2.5333174744172704e-05,
+      "loss": 0.1785,
+      "step": 1423
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.4886702299118042,
+      "learning_rate": 2.5266422852317795e-05,
+      "loss": 0.1388,
+      "step": 1424
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6130447387695312,
+      "learning_rate": 2.5199729274798662e-05,
+      "loss": 0.198,
+      "step": 1425
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6077058911323547,
+      "learning_rate": 2.513309416885865e-05,
+      "loss": 0.2203,
+      "step": 1426
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6400307416915894,
+      "learning_rate": 2.5066517691603324e-05,
+      "loss": 0.1815,
+      "step": 1427
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.49183541536331177,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 0.1378,
+      "step": 1428
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6175490021705627,
+      "learning_rate": 2.493354125087738e-05,
+      "loss": 0.2427,
+      "step": 1429
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5961133241653442,
+      "learning_rate": 2.4867141600925214e-05,
+      "loss": 0.2342,
+      "step": 1430
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.4872584939002991,
+      "learning_rate": 2.4800801206693875e-05,
+      "loss": 0.1899,
+      "step": 1431
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5589633584022522,
+      "learning_rate": 2.4734520224594093e-05,
+      "loss": 0.1932,
+      "step": 1432
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5207188725471497,
+      "learning_rate": 2.4668298810896462e-05,
+      "loss": 0.1742,
+      "step": 1433
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5159749388694763,
+      "learning_rate": 2.4602137121731195e-05,
+      "loss": 0.2069,
+      "step": 1434
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.4745774269104004,
+      "learning_rate": 2.4536035313087602e-05,
+      "loss": 0.1544,
+      "step": 1435
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.46059271693229675,
+      "learning_rate": 2.44699935408139e-05,
+      "loss": 0.1727,
+      "step": 1436
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5216801762580872,
+      "learning_rate": 2.4404011960616747e-05,
+      "loss": 0.1387,
+      "step": 1437
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5417492389678955,
+      "learning_rate": 2.433809072806081e-05,
+      "loss": 0.1746,
+      "step": 1438
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.3360338509082794,
+      "learning_rate": 2.4272229998568575e-05,
+      "loss": 0.0942,
+      "step": 1439
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5473638772964478,
+      "learning_rate": 2.4206429927419794e-05,
+      "loss": 0.1757,
+      "step": 1440
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6473803520202637,
+      "learning_rate": 2.4140690669751283e-05,
+      "loss": 0.2196,
+      "step": 1441
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5746462345123291,
+      "learning_rate": 2.40750123805564e-05,
+      "loss": 0.2242,
+      "step": 1442
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.511986494064331,
+      "learning_rate": 2.4009395214684838e-05,
+      "loss": 0.1446,
+      "step": 1443
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.49034544825553894,
+      "learning_rate": 2.3943839326842092e-05,
+      "loss": 0.1808,
+      "step": 1444
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5102138519287109,
+      "learning_rate": 2.3878344871589257e-05,
+      "loss": 0.1853,
+      "step": 1445
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.46452924609184265,
+      "learning_rate": 2.3812912003342568e-05,
+      "loss": 0.1587,
+      "step": 1446
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.4363616108894348,
+      "learning_rate": 2.3747540876373025e-05,
+      "loss": 0.1584,
+      "step": 1447
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6363255381584167,
+      "learning_rate": 2.3682231644806108e-05,
+      "loss": 0.228,
+      "step": 1448
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6336438059806824,
+      "learning_rate": 2.3616984462621305e-05,
+      "loss": 0.2763,
+      "step": 1449
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.562806248664856,
+      "learning_rate": 2.3551799483651894e-05,
+      "loss": 0.152,
+      "step": 1450
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6084960103034973,
+      "learning_rate": 2.3486676861584405e-05,
+      "loss": 0.1634,
+      "step": 1451
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5762615203857422,
+      "learning_rate": 2.342161674995843e-05,
+      "loss": 0.1668,
+      "step": 1452
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.45737966895103455,
+      "learning_rate": 2.335661930216611e-05,
+      "loss": 0.1444,
+      "step": 1453
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.4499783217906952,
+      "learning_rate": 2.3291684671451903e-05,
+      "loss": 0.1039,
+      "step": 1454
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5764496326446533,
+      "learning_rate": 2.322681301091214e-05,
+      "loss": 0.193,
+      "step": 1455
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5927673578262329,
+      "learning_rate": 2.3162004473494657e-05,
+      "loss": 0.2414,
+      "step": 1456
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.46850934624671936,
+      "learning_rate": 2.3097259211998534e-05,
+      "loss": 0.175,
+      "step": 1457
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.4949398934841156,
+      "learning_rate": 2.3032577379073577e-05,
+      "loss": 0.1646,
+      "step": 1458
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5979020595550537,
+      "learning_rate": 2.296795912722014e-05,
+      "loss": 0.2197,
+      "step": 1459
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.548563539981842,
+      "learning_rate": 2.2903404608788582e-05,
+      "loss": 0.1499,
+      "step": 1460
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.4613582193851471,
+      "learning_rate": 2.283891397597908e-05,
+      "loss": 0.127,
+      "step": 1461
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.6022800803184509,
+      "learning_rate": 2.2774487380841115e-05,
+      "loss": 0.2128,
+      "step": 1462
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5690326690673828,
+      "learning_rate": 2.2710124975273234e-05,
+      "loss": 0.2073,
+      "step": 1463
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5215719938278198,
+      "learning_rate": 2.2645826911022655e-05,
+      "loss": 0.181,
+      "step": 1464
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5557555556297302,
+      "learning_rate": 2.2581593339684835e-05,
+      "loss": 0.1736,
+      "step": 1465
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5484568476676941,
+      "learning_rate": 2.2517424412703252e-05,
+      "loss": 0.1691,
+      "step": 1466
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.4600728154182434,
+      "learning_rate": 2.24533202813689e-05,
+      "loss": 0.1284,
+      "step": 1467
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.3463687002658844,
+      "learning_rate": 2.2389281096820075e-05,
+      "loss": 0.1208,
+      "step": 1468
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.6824270486831665,
+      "learning_rate": 2.2325307010041874e-05,
+      "loss": 0.2289,
+      "step": 1469
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5326138138771057,
+      "learning_rate": 2.2261398171865978e-05,
+      "loss": 0.2068,
+      "step": 1470
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.532956063747406,
+      "learning_rate": 2.21975547329702e-05,
+      "loss": 0.2135,
+      "step": 1471
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5924443006515503,
+      "learning_rate": 2.2133776843878186e-05,
+      "loss": 0.1798,
+      "step": 1472
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6275821328163147,
+      "learning_rate": 2.2070064654958976e-05,
+      "loss": 0.1999,
+      "step": 1473
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5570134520530701,
+      "learning_rate": 2.2006418316426775e-05,
+      "loss": 0.1821,
+      "step": 1474
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.4913628399372101,
+      "learning_rate": 2.1942837978340514e-05,
+      "loss": 0.1524,
+      "step": 1475
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5918798446655273,
+      "learning_rate": 2.187932379060348e-05,
+      "loss": 0.18,
+      "step": 1476
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5336848497390747,
+      "learning_rate": 2.1815875902963055e-05,
+      "loss": 0.1458,
+      "step": 1477
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6554023027420044,
+      "learning_rate": 2.1752494465010242e-05,
+      "loss": 0.2648,
+      "step": 1478
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6325657963752747,
+      "learning_rate": 2.168917962617944e-05,
+      "loss": 0.2216,
+      "step": 1479
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6575515866279602,
+      "learning_rate": 2.1625931535747964e-05,
+      "loss": 0.2249,
+      "step": 1480
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5368912816047668,
+      "learning_rate": 2.1562750342835826e-05,
+      "loss": 0.2033,
+      "step": 1481
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5677066445350647,
+      "learning_rate": 2.1499636196405228e-05,
+      "loss": 0.1763,
+      "step": 1482
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5434936285018921,
+      "learning_rate": 2.1436589245260376e-05,
+      "loss": 0.1458,
+      "step": 1483
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6858835220336914,
+      "learning_rate": 2.1373609638047032e-05,
+      "loss": 0.1865,
+      "step": 1484
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.590842068195343,
+      "learning_rate": 2.1310697523252127e-05,
+      "loss": 0.2181,
+      "step": 1485
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5737771391868591,
+      "learning_rate": 2.1247853049203543e-05,
+      "loss": 0.1513,
+      "step": 1486
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.43269217014312744,
+      "learning_rate": 2.1185076364069617e-05,
+      "loss": 0.1661,
+      "step": 1487
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.4028598368167877,
+      "learning_rate": 2.112236761585892e-05,
+      "loss": 0.11,
+      "step": 1488
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.552455723285675,
+      "learning_rate": 2.1059726952419782e-05,
+      "loss": 0.1704,
+      "step": 1489
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5564739108085632,
+      "learning_rate": 2.09971545214401e-05,
+      "loss": 0.2071,
+      "step": 1490
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6521076560020447,
+      "learning_rate": 2.0934650470446786e-05,
+      "loss": 0.2063,
+      "step": 1491
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6165980696678162,
+      "learning_rate": 2.087221494680563e-05,
+      "loss": 0.2102,
+      "step": 1492
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5333232879638672,
+      "learning_rate": 2.0809848097720826e-05,
+      "loss": 0.1821,
+      "step": 1493
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.6106579303741455,
+      "learning_rate": 2.074755007023461e-05,
+      "loss": 0.1768,
+      "step": 1494
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5013437271118164,
+      "learning_rate": 2.0685321011227036e-05,
+      "loss": 0.1731,
+      "step": 1495
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5261099338531494,
+      "learning_rate": 2.0623161067415463e-05,
+      "loss": 0.1959,
+      "step": 1496
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.4689790904521942,
+      "learning_rate": 2.0561070385354387e-05,
+      "loss": 0.1361,
+      "step": 1497
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.41596245765686035,
+      "learning_rate": 2.049904911143492e-05,
+      "loss": 0.1317,
+      "step": 1498
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.4551275670528412,
+      "learning_rate": 2.0437097391884614e-05,
+      "loss": 0.1239,
+      "step": 1499
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5437151193618774,
+      "learning_rate": 2.0375215372766944e-05,
+      "loss": 0.1719,
+      "step": 1500
+    },
+    {
+      "epoch": 1.42,
+      "eval_loss": 0.2831069529056549,
+      "eval_runtime": 193.5071,
+      "eval_samples_per_second": 1.762,
+      "eval_steps_per_second": 0.222,
+      "step": 1500
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5799771547317505,
+      "learning_rate": 2.0313403199981124e-05,
+      "loss": 0.1788,
+      "step": 1501
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5046328902244568,
+      "learning_rate": 2.025166101926168e-05,
+      "loss": 0.1626,
+      "step": 1502
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5169566869735718,
+      "learning_rate": 2.0189988976178077e-05,
+      "loss": 0.1633,
+      "step": 1503
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7211600542068481,
+      "learning_rate": 2.012838721613447e-05,
+      "loss": 0.2425,
+      "step": 1504
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.625601053237915,
+      "learning_rate": 2.0066855884369244e-05,
+      "loss": 0.1758,
+      "step": 1505
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.670967698097229,
+      "learning_rate": 2.0005395125954812e-05,
+      "loss": 0.2166,
+      "step": 1506
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5635889172554016,
+      "learning_rate": 1.9944005085797124e-05,
+      "loss": 0.1812,
+      "step": 1507
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5200455188751221,
+      "learning_rate": 1.9882685908635462e-05,
+      "loss": 0.212,
+      "step": 1508
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5016717910766602,
+      "learning_rate": 1.982143773904197e-05,
+      "loss": 0.1541,
+      "step": 1509
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5053320527076721,
+      "learning_rate": 1.9760260721421426e-05,
+      "loss": 0.1666,
+      "step": 1510
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5150848031044006,
+      "learning_rate": 1.9699155000010855e-05,
+      "loss": 0.1322,
+      "step": 1511
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5566292405128479,
+      "learning_rate": 1.963812071887913e-05,
+      "loss": 0.1685,
+      "step": 1512
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5469474196434021,
+      "learning_rate": 1.9577158021926774e-05,
+      "loss": 0.1602,
+      "step": 1513
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.654338538646698,
+      "learning_rate": 1.951626705288544e-05,
+      "loss": 0.255,
+      "step": 1514
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.4664451479911804,
+      "learning_rate": 1.9455447955317767e-05,
+      "loss": 0.1539,
+      "step": 1515
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.45495110750198364,
+      "learning_rate": 1.9394700872616855e-05,
+      "loss": 0.0956,
+      "step": 1516
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.5631459355354309,
+      "learning_rate": 1.9334025948006074e-05,
+      "loss": 0.1871,
+      "step": 1517
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.5945963263511658,
+      "learning_rate": 1.9273423324538658e-05,
+      "loss": 0.1986,
+      "step": 1518
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6410242319107056,
+      "learning_rate": 1.9212893145097338e-05,
+      "loss": 0.2174,
+      "step": 1519
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.5024895668029785,
+      "learning_rate": 1.9152435552394104e-05,
+      "loss": 0.1439,
+      "step": 1520
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.5729490518569946,
+      "learning_rate": 1.9092050688969738e-05,
+      "loss": 0.1556,
+      "step": 1521
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.4344686269760132,
+      "learning_rate": 1.903173869719362e-05,
+      "loss": 0.1347,
+      "step": 1522
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.5378993153572083,
+      "learning_rate": 1.8971499719263252e-05,
+      "loss": 0.1812,
+      "step": 1523
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.4713113009929657,
+      "learning_rate": 1.891133389720407e-05,
+      "loss": 0.1566,
+      "step": 1524
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6339928507804871,
+      "learning_rate": 1.8851241372868937e-05,
+      "loss": 0.2246,
+      "step": 1525
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5104354619979858,
+      "learning_rate": 1.8791222287937982e-05,
+      "loss": 0.1349,
+      "step": 1526
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3591233789920807,
+      "learning_rate": 1.873127678391816e-05,
+      "loss": 0.0828,
+      "step": 1527
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5823466777801514,
+      "learning_rate": 1.8671405002142918e-05,
+      "loss": 0.1882,
+      "step": 1528
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5312154293060303,
+      "learning_rate": 1.861160708377193e-05,
+      "loss": 0.1898,
+      "step": 1529
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5790638327598572,
+      "learning_rate": 1.855188316979068e-05,
+      "loss": 0.1577,
+      "step": 1530
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5405199527740479,
+      "learning_rate": 1.8492233401010218e-05,
+      "loss": 0.1795,
+      "step": 1531
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5361924767494202,
+      "learning_rate": 1.8432657918066732e-05,
+      "loss": 0.1842,
+      "step": 1532
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.6220306158065796,
+      "learning_rate": 1.8373156861421325e-05,
+      "loss": 0.232,
+      "step": 1533
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5017096400260925,
+      "learning_rate": 1.831373037135955e-05,
+      "loss": 0.1528,
+      "step": 1534
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5917253494262695,
+      "learning_rate": 1.8254378587991226e-05,
+      "loss": 0.2437,
+      "step": 1535
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.7251100540161133,
+      "learning_rate": 1.8195101651250017e-05,
+      "loss": 0.2391,
+      "step": 1536
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.704756498336792,
+      "learning_rate": 1.8135899700893076e-05,
+      "loss": 0.2225,
+      "step": 1537
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5120024085044861,
+      "learning_rate": 1.807677287650083e-05,
+      "loss": 0.1735,
+      "step": 1538
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.49768704175949097,
+      "learning_rate": 1.8017721317476517e-05,
+      "loss": 0.1431,
+      "step": 1539
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.4535953104496002,
+      "learning_rate": 1.7958745163045986e-05,
+      "loss": 0.131,
+      "step": 1540
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5573491454124451,
+      "learning_rate": 1.7899844552257233e-05,
+      "loss": 0.1584,
+      "step": 1541
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5072364211082458,
+      "learning_rate": 1.7841019623980217e-05,
+      "loss": 0.1423,
+      "step": 1542
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5436108112335205,
+      "learning_rate": 1.778227051690639e-05,
+      "loss": 0.1934,
+      "step": 1543
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5799072980880737,
+      "learning_rate": 1.77235973695485e-05,
+      "loss": 0.2083,
+      "step": 1544
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.47618037462234497,
+      "learning_rate": 1.76650003202402e-05,
+      "loss": 0.1509,
+      "step": 1545
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5910797119140625,
+      "learning_rate": 1.760647950713566e-05,
+      "loss": 0.2049,
+      "step": 1546
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5579193830490112,
+      "learning_rate": 1.75480350682094e-05,
+      "loss": 0.1856,
+      "step": 1547
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.6019135117530823,
+      "learning_rate": 1.74896671412558e-05,
+      "loss": 0.2022,
+      "step": 1548
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.6749143600463867,
+      "learning_rate": 1.7431375863888898e-05,
+      "loss": 0.1875,
+      "step": 1549
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.40615466237068176,
+      "learning_rate": 1.737316137354197e-05,
+      "loss": 0.1156,
+      "step": 1550
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.4887976050376892,
+      "learning_rate": 1.7315023807467295e-05,
+      "loss": 0.1711,
+      "step": 1551
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.6841743588447571,
+      "learning_rate": 1.725696330273575e-05,
+      "loss": 0.2452,
+      "step": 1552
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5371795296669006,
+      "learning_rate": 1.7198979996236546e-05,
+      "loss": 0.1973,
+      "step": 1553
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5669211149215698,
+      "learning_rate": 1.7141074024676912e-05,
+      "loss": 0.2012,
+      "step": 1554
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5980335474014282,
+      "learning_rate": 1.7083245524581666e-05,
+      "loss": 0.2039,
+      "step": 1555
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.6300728917121887,
+      "learning_rate": 1.7025494632293048e-05,
+      "loss": 0.2502,
+      "step": 1556
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.5900150537490845,
+      "learning_rate": 1.6967821483970277e-05,
+      "loss": 0.2144,
+      "step": 1557
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6013487577438354,
+      "learning_rate": 1.6910226215589303e-05,
+      "loss": 0.2069,
+      "step": 1558
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.48544272780418396,
+      "learning_rate": 1.6852708962942427e-05,
+      "loss": 0.1551,
+      "step": 1559
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.7318636178970337,
+      "learning_rate": 1.6795269861638042e-05,
+      "loss": 0.2089,
+      "step": 1560
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.43226975202560425,
+      "learning_rate": 1.6737909047100293e-05,
+      "loss": 0.1311,
+      "step": 1561
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6525475382804871,
+      "learning_rate": 1.6680626654568686e-05,
+      "loss": 0.1945,
+      "step": 1562
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6050831079483032,
+      "learning_rate": 1.6623422819097917e-05,
+      "loss": 0.2058,
+      "step": 1563
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6383846998214722,
+      "learning_rate": 1.656629767555739e-05,
+      "loss": 0.2332,
+      "step": 1564
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.5425280332565308,
+      "learning_rate": 1.650925135863104e-05,
+      "loss": 0.1391,
+      "step": 1565
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6400988698005676,
+      "learning_rate": 1.645228400281689e-05,
+      "loss": 0.2176,
+      "step": 1566
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6015323996543884,
+      "learning_rate": 1.6395395742426874e-05,
+      "loss": 0.1814,
+      "step": 1567
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.5598692893981934,
+      "learning_rate": 1.6338586711586356e-05,
+      "loss": 0.2037,
+      "step": 1568
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.4251973330974579,
+      "learning_rate": 1.6281857044233968e-05,
+      "loss": 0.1173,
+      "step": 1569
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.6665922999382019,
+      "learning_rate": 1.6225206874121218e-05,
+      "loss": 0.2329,
+      "step": 1570
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.5323675870895386,
+      "learning_rate": 1.6168636334812125e-05,
+      "loss": 0.1245,
+      "step": 1571
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.5388573408126831,
+      "learning_rate": 1.6112145559683055e-05,
+      "loss": 0.1549,
+      "step": 1572
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.6083898544311523,
+      "learning_rate": 1.6055734681922223e-05,
+      "loss": 0.2146,
+      "step": 1573
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.46272435784339905,
+      "learning_rate": 1.5999403834529548e-05,
+      "loss": 0.1493,
+      "step": 1574
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.553869366645813,
+      "learning_rate": 1.5943153150316193e-05,
+      "loss": 0.1631,
+      "step": 1575
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.490682452917099,
+      "learning_rate": 1.5886982761904377e-05,
+      "loss": 0.1346,
+      "step": 1576
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.5311291813850403,
+      "learning_rate": 1.583089280172696e-05,
+      "loss": 0.152,
+      "step": 1577
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.4720114469528198,
+      "learning_rate": 1.5774883402027208e-05,
+      "loss": 0.1452,
+      "step": 1578
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5896927118301392,
+      "learning_rate": 1.5718954694858458e-05,
+      "loss": 0.2029,
+      "step": 1579
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5745022296905518,
+      "learning_rate": 1.5663106812083743e-05,
+      "loss": 0.183,
+      "step": 1580
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6028936505317688,
+      "learning_rate": 1.5607339885375617e-05,
+      "loss": 0.2097,
+      "step": 1581
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.594720184803009,
+      "learning_rate": 1.555165404621567e-05,
+      "loss": 0.2284,
+      "step": 1582
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5002874732017517,
+      "learning_rate": 1.549604942589441e-05,
+      "loss": 0.1156,
+      "step": 1583
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5099266767501831,
+      "learning_rate": 1.5440526155510763e-05,
+      "loss": 0.2049,
+      "step": 1584
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6339079737663269,
+      "learning_rate": 1.5385084365971948e-05,
+      "loss": 0.213,
+      "step": 1585
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5323735475540161,
+      "learning_rate": 1.532972418799298e-05,
+      "loss": 0.1954,
+      "step": 1586
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5360701084136963,
+      "learning_rate": 1.527444575209654e-05,
+      "loss": 0.173,
+      "step": 1587
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5458484292030334,
+      "learning_rate": 1.5219249188612556e-05,
+      "loss": 0.2039,
+      "step": 1588
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.4940440356731415,
+      "learning_rate": 1.5164134627677895e-05,
+      "loss": 0.188,
+      "step": 1589
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5832428336143494,
+      "learning_rate": 1.5109102199236152e-05,
+      "loss": 0.206,
+      "step": 1590
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5490756034851074,
+      "learning_rate": 1.5054152033037205e-05,
+      "loss": 0.1873,
+      "step": 1591
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.42287158966064453,
+      "learning_rate": 1.4999284258637053e-05,
+      "loss": 0.1387,
+      "step": 1592
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5771350860595703,
+      "learning_rate": 1.4944499005397371e-05,
+      "loss": 0.2236,
+      "step": 1593
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5058189034461975,
+      "learning_rate": 1.488979640248534e-05,
+      "loss": 0.1772,
+      "step": 1594
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.6315512657165527,
+      "learning_rate": 1.4835176578873212e-05,
+      "loss": 0.1779,
+      "step": 1595
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.6534866690635681,
+      "learning_rate": 1.4780639663338125e-05,
+      "loss": 0.2096,
+      "step": 1596
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.563092052936554,
+      "learning_rate": 1.4726185784461728e-05,
+      "loss": 0.1735,
+      "step": 1597
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.3996272087097168,
+      "learning_rate": 1.4671815070629868e-05,
+      "loss": 0.11,
+      "step": 1598
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.565413236618042,
+      "learning_rate": 1.461752765003236e-05,
+      "loss": 0.2126,
+      "step": 1599
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.6139777302742004,
+      "learning_rate": 1.4563323650662586e-05,
+      "loss": 0.2163,
+      "step": 1600
+    },
+    {
+      "epoch": 1.52,
+      "eval_loss": 0.28096804022789,
+      "eval_runtime": 195.0677,
+      "eval_samples_per_second": 1.748,
+      "eval_steps_per_second": 0.22,
+      "step": 1600
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.617189347743988,
+      "learning_rate": 1.45092032003173e-05,
+      "loss": 0.2364,
+      "step": 1601
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.6501761674880981,
+      "learning_rate": 1.4455166426596223e-05,
+      "loss": 0.2473,
+      "step": 1602
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5644694566726685,
+      "learning_rate": 1.4401213456901819e-05,
+      "loss": 0.1397,
+      "step": 1603
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.47821709513664246,
+      "learning_rate": 1.434734441843899e-05,
+      "loss": 0.1376,
+      "step": 1604
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.6071019768714905,
+      "learning_rate": 1.4293559438214688e-05,
+      "loss": 0.2072,
+      "step": 1605
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5405787825584412,
+      "learning_rate": 1.4239858643037751e-05,
+      "loss": 0.2025,
+      "step": 1606
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5375504493713379,
+      "learning_rate": 1.4186242159518476e-05,
+      "loss": 0.1874,
+      "step": 1607
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5260862112045288,
+      "learning_rate": 1.4132710114068426e-05,
+      "loss": 0.1501,
+      "step": 1608
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.565697431564331,
+      "learning_rate": 1.4079262632900047e-05,
+      "loss": 0.2064,
+      "step": 1609
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5254867076873779,
+      "learning_rate": 1.4025899842026442e-05,
+      "loss": 0.1559,
+      "step": 1610
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.4235105812549591,
+      "learning_rate": 1.3972621867260998e-05,
+      "loss": 0.1417,
+      "step": 1611
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5646243095397949,
+      "learning_rate": 1.3919428834217163e-05,
+      "loss": 0.1824,
+      "step": 1612
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5898502469062805,
+      "learning_rate": 1.3866320868308136e-05,
+      "loss": 0.1543,
+      "step": 1613
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.4921277165412903,
+      "learning_rate": 1.3813298094746491e-05,
+      "loss": 0.1555,
+      "step": 1614
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5271590352058411,
+      "learning_rate": 1.3760360638544012e-05,
+      "loss": 0.1858,
+      "step": 1615
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5243542194366455,
+      "learning_rate": 1.3707508624511262e-05,
+      "loss": 0.171,
+      "step": 1616
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6124130487442017,
+      "learning_rate": 1.3654742177257434e-05,
+      "loss": 0.2198,
+      "step": 1617
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6262127161026001,
+      "learning_rate": 1.36020614211899e-05,
+      "loss": 0.1922,
+      "step": 1618
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.587662398815155,
+      "learning_rate": 1.3549466480514078e-05,
+      "loss": 0.2327,
+      "step": 1619
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.43168455362319946,
+      "learning_rate": 1.349695747923298e-05,
+      "loss": 0.1536,
+      "step": 1620
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.4846075773239136,
+      "learning_rate": 1.3444534541147057e-05,
+      "loss": 0.1165,
+      "step": 1621
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5710000395774841,
+      "learning_rate": 1.339219778985385e-05,
+      "loss": 0.1649,
+      "step": 1622
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5888062119483948,
+      "learning_rate": 1.3339947348747633e-05,
+      "loss": 0.1788,
+      "step": 1623
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.4962611496448517,
+      "learning_rate": 1.3287783341019278e-05,
+      "loss": 0.1495,
+      "step": 1624
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.47519540786743164,
+      "learning_rate": 1.323570588965578e-05,
+      "loss": 0.157,
+      "step": 1625
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.48729145526885986,
+      "learning_rate": 1.3183715117440142e-05,
+      "loss": 0.1346,
+      "step": 1626
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.4600526690483093,
+      "learning_rate": 1.3131811146950945e-05,
+      "loss": 0.1545,
+      "step": 1627
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5114222168922424,
+      "learning_rate": 1.3079994100562159e-05,
+      "loss": 0.1637,
+      "step": 1628
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5812637209892273,
+      "learning_rate": 1.3028264100442771e-05,
+      "loss": 0.2038,
+      "step": 1629
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.4426330029964447,
+      "learning_rate": 1.2976621268556571e-05,
+      "loss": 0.1313,
+      "step": 1630
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.45024406909942627,
+      "learning_rate": 1.2925065726661844e-05,
+      "loss": 0.148,
+      "step": 1631
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5679466128349304,
+      "learning_rate": 1.2873597596311027e-05,
+      "loss": 0.1959,
+      "step": 1632
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.41410356760025024,
+      "learning_rate": 1.2822216998850506e-05,
+      "loss": 0.1127,
+      "step": 1633
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5192013382911682,
+      "learning_rate": 1.2770924055420258e-05,
+      "loss": 0.1714,
+      "step": 1634
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.49232709407806396,
+      "learning_rate": 1.2719718886953646e-05,
+      "loss": 0.1436,
+      "step": 1635
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.4858894944190979,
+      "learning_rate": 1.2668601614177017e-05,
+      "loss": 0.1313,
+      "step": 1636
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6338505744934082,
+      "learning_rate": 1.2617572357609564e-05,
+      "loss": 0.253,
+      "step": 1637
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6390866637229919,
+      "learning_rate": 1.2566631237562893e-05,
+      "loss": 0.2129,
+      "step": 1638
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6305318474769592,
+      "learning_rate": 1.2515778374140858e-05,
+      "loss": 0.2609,
+      "step": 1639
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5560963749885559,
+      "learning_rate": 1.246501388723923e-05,
+      "loss": 0.1874,
+      "step": 1640
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5227592587471008,
+      "learning_rate": 1.2414337896545374e-05,
+      "loss": 0.187,
+      "step": 1641
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.4331105053424835,
+      "learning_rate": 1.2363750521538064e-05,
+      "loss": 0.0882,
+      "step": 1642
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.639569878578186,
+      "learning_rate": 1.2313251881487082e-05,
+      "loss": 0.2478,
+      "step": 1643
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5105839371681213,
+      "learning_rate": 1.2262842095453065e-05,
+      "loss": 0.1509,
+      "step": 1644
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5753269791603088,
+      "learning_rate": 1.2212521282287092e-05,
+      "loss": 0.1767,
+      "step": 1645
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5563485026359558,
+      "learning_rate": 1.2162289560630524e-05,
+      "loss": 0.2097,
+      "step": 1646
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5199491381645203,
+      "learning_rate": 1.2112147048914668e-05,
+      "loss": 0.1869,
+      "step": 1647
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.4894522726535797,
+      "learning_rate": 1.2062093865360458e-05,
+      "loss": 0.137,
+      "step": 1648
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5169162750244141,
+      "learning_rate": 1.2012130127978266e-05,
+      "loss": 0.1612,
+      "step": 1649
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.47476503252983093,
+      "learning_rate": 1.1962255954567537e-05,
+      "loss": 0.1724,
+      "step": 1650
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5074109435081482,
+      "learning_rate": 1.1912471462716595e-05,
+      "loss": 0.147,
+      "step": 1651
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.4853825271129608,
+      "learning_rate": 1.1862776769802276e-05,
+      "loss": 0.1345,
+      "step": 1652
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3899320960044861,
+      "learning_rate": 1.1813171992989741e-05,
+      "loss": 0.1206,
+      "step": 1653
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.4748738706111908,
+      "learning_rate": 1.1763657249232107e-05,
+      "loss": 0.1332,
+      "step": 1654
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5740524530410767,
+      "learning_rate": 1.1714232655270263e-05,
+      "loss": 0.1951,
+      "step": 1655
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.48167064785957336,
+      "learning_rate": 1.1664898327632551e-05,
+      "loss": 0.1276,
+      "step": 1656
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.52054363489151,
+      "learning_rate": 1.1615654382634444e-05,
+      "loss": 0.1425,
+      "step": 1657
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.46148741245269775,
+      "learning_rate": 1.1566500936378388e-05,
+      "loss": 0.1324,
+      "step": 1658
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.6298129558563232,
+      "learning_rate": 1.1517438104753386e-05,
+      "loss": 0.2103,
+      "step": 1659
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.49577072262763977,
+      "learning_rate": 1.146846600343488e-05,
+      "loss": 0.149,
+      "step": 1660
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.45052483677864075,
+      "learning_rate": 1.1419584747884322e-05,
+      "loss": 0.1512,
+      "step": 1661
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5833991169929504,
+      "learning_rate": 1.1370794453349037e-05,
+      "loss": 0.1355,
+      "step": 1662
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.5447257161140442,
+      "learning_rate": 1.132209523486184e-05,
+      "loss": 0.1862,
+      "step": 1663
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.3919275403022766,
+      "learning_rate": 1.1273487207240847e-05,
+      "loss": 0.1216,
+      "step": 1664
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.5174094438552856,
+      "learning_rate": 1.1224970485089192e-05,
+      "loss": 0.1705,
+      "step": 1665
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.660071074962616,
+      "learning_rate": 1.1176545182794674e-05,
+      "loss": 0.213,
+      "step": 1666
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.6224027276039124,
+      "learning_rate": 1.1128211414529627e-05,
+      "loss": 0.236,
+      "step": 1667
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.6973505616188049,
+      "learning_rate": 1.1079969294250514e-05,
+      "loss": 0.2229,
+      "step": 1668
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.5637111067771912,
+      "learning_rate": 1.1031818935697763e-05,
+      "loss": 0.1623,
+      "step": 1669
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4882502257823944,
+      "learning_rate": 1.0983760452395413e-05,
+      "loss": 0.1438,
+      "step": 1670
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.43866851925849915,
+      "learning_rate": 1.0935793957650948e-05,
+      "loss": 0.114,
+      "step": 1671
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.3543342053890228,
+      "learning_rate": 1.0887919564554894e-05,
+      "loss": 0.0951,
+      "step": 1672
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.5770727396011353,
+      "learning_rate": 1.0840137385980697e-05,
+      "loss": 0.1799,
+      "step": 1673
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4266512393951416,
+      "learning_rate": 1.079244753458437e-05,
+      "loss": 0.1263,
+      "step": 1674
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.5762373805046082,
+      "learning_rate": 1.0744850122804218e-05,
+      "loss": 0.2194,
+      "step": 1675
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.5406520962715149,
+      "learning_rate": 1.0697345262860636e-05,
+      "loss": 0.2057,
+      "step": 1676
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6332505345344543,
+      "learning_rate": 1.064993306675578e-05,
+      "loss": 0.2006,
+      "step": 1677
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.3984108567237854,
+      "learning_rate": 1.0602613646273374e-05,
+      "loss": 0.1317,
+      "step": 1678
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6551477313041687,
+      "learning_rate": 1.0555387112978349e-05,
+      "loss": 0.1657,
+      "step": 1679
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.5520603656768799,
+      "learning_rate": 1.0508253578216693e-05,
+      "loss": 0.18,
+      "step": 1680
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.543144941329956,
+      "learning_rate": 1.046121315311508e-05,
+      "loss": 0.1683,
+      "step": 1681
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4693453311920166,
+      "learning_rate": 1.0414265948580692e-05,
+      "loss": 0.1789,
+      "step": 1682
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.5766890048980713,
+      "learning_rate": 1.0367412075300942e-05,
+      "loss": 0.2091,
+      "step": 1683
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5147340297698975,
+      "learning_rate": 1.032065164374313e-05,
+      "loss": 0.1435,
+      "step": 1684
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5613285899162292,
+      "learning_rate": 1.0273984764154326e-05,
+      "loss": 0.1508,
+      "step": 1685
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.6336666345596313,
+      "learning_rate": 1.0227411546560962e-05,
+      "loss": 0.2145,
+      "step": 1686
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.4728489816188812,
+      "learning_rate": 1.0180932100768715e-05,
+      "loss": 0.1723,
+      "step": 1687
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.7171174883842468,
+      "learning_rate": 1.01345465363621e-05,
+      "loss": 0.216,
+      "step": 1688
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5459414720535278,
+      "learning_rate": 1.0088254962704341e-05,
+      "loss": 0.2187,
+      "step": 1689
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.47307372093200684,
+      "learning_rate": 1.0042057488937067e-05,
+      "loss": 0.1304,
+      "step": 1690
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5610155463218689,
+      "learning_rate": 9.995954223979992e-06,
+      "loss": 0.1608,
+      "step": 1691
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5830497145652771,
+      "learning_rate": 9.949945276530781e-06,
+      "loss": 0.1829,
+      "step": 1692
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.6063210964202881,
+      "learning_rate": 9.904030755064659e-06,
+      "loss": 0.186,
+      "step": 1693
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5668680667877197,
+      "learning_rate": 9.858210767834292e-06,
+      "loss": 0.1737,
+      "step": 1694
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.564145028591156,
+      "learning_rate": 9.8124854228694e-06,
+      "loss": 0.1559,
+      "step": 1695
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5226021409034729,
+      "learning_rate": 9.766854827976617e-06,
+      "loss": 0.1721,
+      "step": 1696
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.4931643605232239,
+      "learning_rate": 9.72131909073914e-06,
+      "loss": 0.1436,
+      "step": 1697
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.6431955099105835,
+      "learning_rate": 9.675878318516546e-06,
+      "loss": 0.2084,
+      "step": 1698
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5106863975524902,
+      "learning_rate": 9.630532618444533e-06,
+      "loss": 0.1764,
+      "step": 1699
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5867109298706055,
+      "learning_rate": 9.58528209743459e-06,
+      "loss": 0.2047,
+      "step": 1700
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 0.28330597281455994,
+      "eval_runtime": 194.4963,
+      "eval_samples_per_second": 1.753,
+      "eval_steps_per_second": 0.221,
+      "step": 1700
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.44337424635887146,
+      "learning_rate": 9.540126862173864e-06,
+      "loss": 0.1434,
+      "step": 1701
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5499069690704346,
+      "learning_rate": 9.495067019124792e-06,
+      "loss": 0.1717,
+      "step": 1702
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5205509662628174,
+      "learning_rate": 9.450102674524952e-06,
+      "loss": 0.1629,
+      "step": 1703
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5879272222518921,
+      "learning_rate": 9.405233934386726e-06,
+      "loss": 0.2129,
+      "step": 1704
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.556326150894165,
+      "learning_rate": 9.360460904497131e-06,
+      "loss": 0.1862,
+      "step": 1705
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.6327759027481079,
+      "learning_rate": 9.315783690417479e-06,
+      "loss": 0.2371,
+      "step": 1706
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.7272502183914185,
+      "learning_rate": 9.271202397483215e-06,
+      "loss": 0.1805,
+      "step": 1707
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5610241889953613,
+      "learning_rate": 9.226717130803636e-06,
+      "loss": 0.1736,
+      "step": 1708
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5325340032577515,
+      "learning_rate": 9.182327995261591e-06,
+      "loss": 0.1885,
+      "step": 1709
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.418314129114151,
+      "learning_rate": 9.138035095513336e-06,
+      "loss": 0.1145,
+      "step": 1710
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.6141099333763123,
+      "learning_rate": 9.09383853598818e-06,
+      "loss": 0.1848,
+      "step": 1711
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5269084572792053,
+      "learning_rate": 9.049738420888349e-06,
+      "loss": 0.1712,
+      "step": 1712
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.6486226916313171,
+      "learning_rate": 9.005734854188625e-06,
+      "loss": 0.2275,
+      "step": 1713
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5812177658081055,
+      "learning_rate": 8.961827939636196e-06,
+      "loss": 0.1768,
+      "step": 1714
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.5436131954193115,
+      "learning_rate": 8.918017780750349e-06,
+      "loss": 0.1826,
+      "step": 1715
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.55128413438797,
+      "learning_rate": 8.87430448082227e-06,
+      "loss": 0.1588,
+      "step": 1716
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.6647340655326843,
+      "learning_rate": 8.830688142914784e-06,
+      "loss": 0.2368,
+      "step": 1717
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.4678126573562622,
+      "learning_rate": 8.787168869862066e-06,
+      "loss": 0.1551,
+      "step": 1718
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.5900838375091553,
+      "learning_rate": 8.743746764269512e-06,
+      "loss": 0.1915,
+      "step": 1719
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.581432580947876,
+      "learning_rate": 8.700421928513352e-06,
+      "loss": 0.2246,
+      "step": 1720
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.5566329956054688,
+      "learning_rate": 8.657194464740542e-06,
+      "loss": 0.1737,
+      "step": 1721
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.7829611301422119,
+      "learning_rate": 8.614064474868421e-06,
+      "loss": 0.2747,
+      "step": 1722
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.5121700763702393,
+      "learning_rate": 8.571032060584556e-06,
+      "loss": 0.1629,
+      "step": 1723
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.6434698104858398,
+      "learning_rate": 8.528097323346407e-06,
+      "loss": 0.18,
+      "step": 1724
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.6498457193374634,
+      "learning_rate": 8.485260364381187e-06,
+      "loss": 0.2326,
+      "step": 1725
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5657123923301697,
+      "learning_rate": 8.442521284685573e-06,
+      "loss": 0.181,
+      "step": 1726
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.6552181839942932,
+      "learning_rate": 8.399880185025438e-06,
+      "loss": 0.2184,
+      "step": 1727
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.6922997832298279,
+      "learning_rate": 8.357337165935675e-06,
+      "loss": 0.243,
+      "step": 1728
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5678548812866211,
+      "learning_rate": 8.314892327719936e-06,
+      "loss": 0.237,
+      "step": 1729
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5033883452415466,
+      "learning_rate": 8.27254577045039e-06,
+      "loss": 0.166,
+      "step": 1730
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.577303946018219,
+      "learning_rate": 8.230297593967461e-06,
+      "loss": 0.1809,
+      "step": 1731
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.535258948802948,
+      "learning_rate": 8.188147897879667e-06,
+      "loss": 0.1744,
+      "step": 1732
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5396235585212708,
+      "learning_rate": 8.146096781563284e-06,
+      "loss": 0.1949,
+      "step": 1733
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.48605915904045105,
+      "learning_rate": 8.104144344162229e-06,
+      "loss": 0.1525,
+      "step": 1734
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.49206918478012085,
+      "learning_rate": 8.062290684587698e-06,
+      "loss": 0.1531,
+      "step": 1735
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.42672696709632874,
+      "learning_rate": 8.02053590151805e-06,
+      "loss": 0.1239,
+      "step": 1736
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.4577820599079132,
+      "learning_rate": 7.978880093398516e-06,
+      "loss": 0.1338,
+      "step": 1737
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.38631829619407654,
+      "learning_rate": 7.937323358440935e-06,
+      "loss": 0.0851,
+      "step": 1738
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5847574472427368,
+      "learning_rate": 7.89586579462362e-06,
+      "loss": 0.1876,
+      "step": 1739
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.46370571851730347,
+      "learning_rate": 7.854507499691005e-06,
+      "loss": 0.1409,
+      "step": 1740
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5370524525642395,
+      "learning_rate": 7.813248571153542e-06,
+      "loss": 0.1836,
+      "step": 1741
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5397897362709045,
+      "learning_rate": 7.772089106287344e-06,
+      "loss": 0.1977,
+      "step": 1742
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5646621584892273,
+      "learning_rate": 7.731029202134078e-06,
+      "loss": 0.1483,
+      "step": 1743
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5109905004501343,
+      "learning_rate": 7.690068955500624e-06,
+      "loss": 0.1528,
+      "step": 1744
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5900869369506836,
+      "learning_rate": 7.649208462958935e-06,
+      "loss": 0.205,
+      "step": 1745
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5813480615615845,
+      "learning_rate": 7.608447820845771e-06,
+      "loss": 0.1916,
+      "step": 1746
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5547435283660889,
+      "learning_rate": 7.5677871252624485e-06,
+      "loss": 0.1857,
+      "step": 1747
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5200846791267395,
+      "learning_rate": 7.527226472074678e-06,
+      "loss": 0.1766,
+      "step": 1748
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.587592601776123,
+      "learning_rate": 7.486765956912262e-06,
+      "loss": 0.1901,
+      "step": 1749
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5553120970726013,
+      "learning_rate": 7.446405675168938e-06,
+      "loss": 0.1477,
+      "step": 1750
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.6246035695075989,
+      "learning_rate": 7.4061457220021e-06,
+      "loss": 0.2408,
+      "step": 1751
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5805262923240662,
+      "learning_rate": 7.365986192332625e-06,
+      "loss": 0.1569,
+      "step": 1752
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.6110222339630127,
+      "learning_rate": 7.325927180844589e-06,
+      "loss": 0.1883,
+      "step": 1753
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5690113306045532,
+      "learning_rate": 7.285968781985092e-06,
+      "loss": 0.1652,
+      "step": 1754
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.6819297671318054,
+      "learning_rate": 7.246111089964042e-06,
+      "loss": 0.2272,
+      "step": 1755
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5570441484451294,
+      "learning_rate": 7.206354198753862e-06,
+      "loss": 0.1576,
+      "step": 1756
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.4810102880001068,
+      "learning_rate": 7.166698202089367e-06,
+      "loss": 0.144,
+      "step": 1757
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5559511780738831,
+      "learning_rate": 7.127143193467445e-06,
+      "loss": 0.2026,
+      "step": 1758
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.4678940773010254,
+      "learning_rate": 7.087689266146935e-06,
+      "loss": 0.1463,
+      "step": 1759
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5506646633148193,
+      "learning_rate": 7.048336513148307e-06,
+      "loss": 0.1891,
+      "step": 1760
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5009552836418152,
+      "learning_rate": 7.009085027253543e-06,
+      "loss": 0.1524,
+      "step": 1761
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.4613531231880188,
+      "learning_rate": 6.969934901005809e-06,
+      "loss": 0.1257,
+      "step": 1762
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5689277052879333,
+      "learning_rate": 6.930886226709343e-06,
+      "loss": 0.2051,
+      "step": 1763
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5405222177505493,
+      "learning_rate": 6.8919390964291895e-06,
+      "loss": 0.186,
+      "step": 1764
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5631533861160278,
+      "learning_rate": 6.853093601990945e-06,
+      "loss": 0.1986,
+      "step": 1765
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5106444954872131,
+      "learning_rate": 6.814349834980621e-06,
+      "loss": 0.1774,
+      "step": 1766
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.4936462938785553,
+      "learning_rate": 6.775707886744342e-06,
+      "loss": 0.1162,
+      "step": 1767
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.6819619536399841,
+      "learning_rate": 6.7371678483882264e-06,
+      "loss": 0.2273,
+      "step": 1768
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.6219656467437744,
+      "learning_rate": 6.698729810778065e-06,
+      "loss": 0.2017,
+      "step": 1769
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5906422138214111,
+      "learning_rate": 6.660393864539222e-06,
+      "loss": 0.1996,
+      "step": 1770
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.45882487297058105,
+      "learning_rate": 6.622160100056296e-06,
+      "loss": 0.1312,
+      "step": 1771
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.4272960126399994,
+      "learning_rate": 6.58402860747302e-06,
+      "loss": 0.1392,
+      "step": 1772
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.4796423316001892,
+      "learning_rate": 6.5459994766919945e-06,
+      "loss": 0.1484,
+      "step": 1773
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.7045693397521973,
+      "learning_rate": 6.508072797374454e-06,
+      "loss": 0.2761,
+      "step": 1774
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5578077435493469,
+      "learning_rate": 6.470248658940115e-06,
+      "loss": 0.1845,
+      "step": 1775
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5014683604240417,
+      "learning_rate": 6.4325271505669025e-06,
+      "loss": 0.1534,
+      "step": 1776
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.7994627952575684,
+      "learning_rate": 6.394908361190805e-06,
+      "loss": 0.2254,
+      "step": 1777
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5447471737861633,
+      "learning_rate": 6.3573923795056e-06,
+      "loss": 0.1664,
+      "step": 1778
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5928052067756653,
+      "learning_rate": 6.319979293962691e-06,
+      "loss": 0.2328,
+      "step": 1779
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.45398178696632385,
+      "learning_rate": 6.282669192770896e-06,
+      "loss": 0.1376,
+      "step": 1780
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.4510864317417145,
+      "learning_rate": 6.245462163896188e-06,
+      "loss": 0.1223,
+      "step": 1781
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5491384267807007,
+      "learning_rate": 6.208358295061572e-06,
+      "loss": 0.1723,
+      "step": 1782
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.6069249510765076,
+      "learning_rate": 6.171357673746797e-06,
+      "loss": 0.188,
+      "step": 1783
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5465146899223328,
+      "learning_rate": 6.134460387188206e-06,
+      "loss": 0.2043,
+      "step": 1784
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5597583055496216,
+      "learning_rate": 6.097666522378498e-06,
+      "loss": 0.1794,
+      "step": 1785
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5075501203536987,
+      "learning_rate": 6.060976166066546e-06,
+      "loss": 0.1273,
+      "step": 1786
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.608333945274353,
+      "learning_rate": 6.024389404757164e-06,
+      "loss": 0.2198,
+      "step": 1787
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.4696008563041687,
+      "learning_rate": 5.9879063247109405e-06,
+      "loss": 0.1492,
+      "step": 1788
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6471996903419495,
+      "learning_rate": 5.951527011944008e-06,
+      "loss": 0.1738,
+      "step": 1789
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6312731504440308,
+      "learning_rate": 5.915251552227829e-06,
+      "loss": 0.2091,
+      "step": 1790
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5242721438407898,
+      "learning_rate": 5.879080031089046e-06,
+      "loss": 0.1938,
+      "step": 1791
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.571564793586731,
+      "learning_rate": 5.8430125338092115e-06,
+      "loss": 0.1683,
+      "step": 1792
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.722872257232666,
+      "learning_rate": 5.807049145424648e-06,
+      "loss": 0.2329,
+      "step": 1793
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5407509207725525,
+      "learning_rate": 5.771189950726191e-06,
+      "loss": 0.1317,
+      "step": 1794
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.567973792552948,
+      "learning_rate": 5.7354350342590566e-06,
+      "loss": 0.1948,
+      "step": 1795
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5687444806098938,
+      "learning_rate": 5.699784480322568e-06,
+      "loss": 0.1689,
+      "step": 1796
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.596442699432373,
+      "learning_rate": 5.664238372970015e-06,
+      "loss": 0.2028,
+      "step": 1797
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6144737005233765,
+      "learning_rate": 5.628796796008434e-06,
+      "loss": 0.1738,
+      "step": 1798
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6136963963508606,
+      "learning_rate": 5.5934598329983874e-06,
+      "loss": 0.1696,
+      "step": 1799
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.44626152515411377,
+      "learning_rate": 5.558227567253832e-06,
+      "loss": 0.1123,
+      "step": 1800
+    },
+    {
+      "epoch": 1.71,
+      "eval_loss": 0.28163015842437744,
+      "eval_runtime": 193.8712,
+      "eval_samples_per_second": 1.759,
+      "eval_steps_per_second": 0.222,
+      "step": 1800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.8069314956665039,
+      "learning_rate": 5.52310008184182e-06,
+      "loss": 0.1949,
+      "step": 1801
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.56414794921875,
+      "learning_rate": 5.488077459582425e-06,
+      "loss": 0.1886,
+      "step": 1802
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.49857449531555176,
+      "learning_rate": 5.453159783048433e-06,
+      "loss": 0.1558,
+      "step": 1803
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5731927156448364,
+      "learning_rate": 5.418347134565249e-06,
+      "loss": 0.1871,
+      "step": 1804
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.8625755310058594,
+      "learning_rate": 5.383639596210604e-06,
+      "loss": 0.29,
+      "step": 1805
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5441470742225647,
+      "learning_rate": 5.349037249814442e-06,
+      "loss": 0.1878,
+      "step": 1806
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.6075496673583984,
+      "learning_rate": 5.314540176958699e-06,
+      "loss": 0.1998,
+      "step": 1807
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5445616841316223,
+      "learning_rate": 5.28014845897708e-06,
+      "loss": 0.2039,
+      "step": 1808
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.4646284580230713,
+      "learning_rate": 5.24586217695493e-06,
+      "loss": 0.1544,
+      "step": 1809
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.583258330821991,
+      "learning_rate": 5.211681411728969e-06,
+      "loss": 0.203,
+      "step": 1810
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.3626076579093933,
+      "learning_rate": 5.177606243887184e-06,
+      "loss": 0.0605,
+      "step": 1811
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.47434619069099426,
+      "learning_rate": 5.143636753768549e-06,
+      "loss": 0.1638,
+      "step": 1812
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.47253933548927307,
+      "learning_rate": 5.109773021462921e-06,
+      "loss": 0.1351,
+      "step": 1813
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.5626814365386963,
+      "learning_rate": 5.076015126810784e-06,
+      "loss": 0.2134,
+      "step": 1814
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4918833374977112,
+      "learning_rate": 5.042363149403107e-06,
+      "loss": 0.1333,
+      "step": 1815
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.6245091557502747,
+      "learning_rate": 5.008817168581137e-06,
+      "loss": 0.2291,
+      "step": 1816
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.5398136377334595,
+      "learning_rate": 4.975377263436193e-06,
+      "loss": 0.166,
+      "step": 1817
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.5186338424682617,
+      "learning_rate": 4.94204351280953e-06,
+      "loss": 0.1692,
+      "step": 1818
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.5945393443107605,
+      "learning_rate": 4.908815995292082e-06,
+      "loss": 0.2151,
+      "step": 1819
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.48832395672798157,
+      "learning_rate": 4.8756947892243725e-06,
+      "loss": 0.1079,
+      "step": 1820
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5624931454658508,
+      "learning_rate": 4.842679972696213e-06,
+      "loss": 0.1805,
+      "step": 1821
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5417201519012451,
+      "learning_rate": 4.809771623546627e-06,
+      "loss": 0.1785,
+      "step": 1822
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.7733779549598694,
+      "learning_rate": 4.7769698193636135e-06,
+      "loss": 0.3127,
+      "step": 1823
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.45463699102401733,
+      "learning_rate": 4.744274637483936e-06,
+      "loss": 0.1344,
+      "step": 1824
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.6118283867835999,
+      "learning_rate": 4.711686154993028e-06,
+      "loss": 0.172,
+      "step": 1825
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5231435298919678,
+      "learning_rate": 4.6792044487247e-06,
+      "loss": 0.1138,
+      "step": 1826
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.49529483914375305,
+      "learning_rate": 4.646829595261071e-06,
+      "loss": 0.1385,
+      "step": 1827
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.589891254901886,
+      "learning_rate": 4.614561670932288e-06,
+      "loss": 0.1934,
+      "step": 1828
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5054509043693542,
+      "learning_rate": 4.582400751816435e-06,
+      "loss": 0.1474,
+      "step": 1829
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.48150351643562317,
+      "learning_rate": 4.550346913739256e-06,
+      "loss": 0.1482,
+      "step": 1830
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5984383225440979,
+      "learning_rate": 4.5184002322740785e-06,
+      "loss": 0.139,
+      "step": 1831
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.5875803232192993,
+      "learning_rate": 4.486560782741578e-06,
+      "loss": 0.1585,
+      "step": 1832
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.572141706943512,
+      "learning_rate": 4.454828640209574e-06,
+      "loss": 0.1981,
+      "step": 1833
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.6422932744026184,
+      "learning_rate": 4.423203879492943e-06,
+      "loss": 0.2216,
+      "step": 1834
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.5529267191886902,
+      "learning_rate": 4.391686575153331e-06,
+      "loss": 0.2022,
+      "step": 1835
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.5153523683547974,
+      "learning_rate": 4.360276801499086e-06,
+      "loss": 0.1437,
+      "step": 1836
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.4966295063495636,
+      "learning_rate": 4.328974632584992e-06,
+      "loss": 0.1625,
+      "step": 1837
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.4619586169719696,
+      "learning_rate": 4.29778014221216e-06,
+      "loss": 0.1464,
+      "step": 1838
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.5502638220787048,
+      "learning_rate": 4.266693403927802e-06,
+      "loss": 0.1488,
+      "step": 1839
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.5717570185661316,
+      "learning_rate": 4.2357144910251e-06,
+      "loss": 0.1809,
+      "step": 1840
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.617010772228241,
+      "learning_rate": 4.20484347654303e-06,
+      "loss": 0.2514,
+      "step": 1841
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.48656606674194336,
+      "learning_rate": 4.174080433266136e-06,
+      "loss": 0.1382,
+      "step": 1842
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.39796069264411926,
+      "learning_rate": 4.1434254337244404e-06,
+      "loss": 0.1246,
+      "step": 1843
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.6167749166488647,
+      "learning_rate": 4.112878550193194e-06,
+      "loss": 0.1683,
+      "step": 1844
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5133748650550842,
+      "learning_rate": 4.082439854692782e-06,
+      "loss": 0.1223,
+      "step": 1845
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5869962573051453,
+      "learning_rate": 4.05210941898847e-06,
+      "loss": 0.1944,
+      "step": 1846
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5568535327911377,
+      "learning_rate": 4.021887314590323e-06,
+      "loss": 0.1529,
+      "step": 1847
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5962230563163757,
+      "learning_rate": 3.991773612752952e-06,
+      "loss": 0.1856,
+      "step": 1848
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.4952459931373596,
+      "learning_rate": 3.961768384475428e-06,
+      "loss": 0.1622,
+      "step": 1849
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5177804231643677,
+      "learning_rate": 3.9318717005010495e-06,
+      "loss": 0.1739,
+      "step": 1850
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5251696109771729,
+      "learning_rate": 3.902083631317193e-06,
+      "loss": 0.1513,
+      "step": 1851
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.4898911714553833,
+      "learning_rate": 3.872404247155193e-06,
+      "loss": 0.1628,
+      "step": 1852
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.4214845597743988,
+      "learning_rate": 3.842833617990077e-06,
+      "loss": 0.1332,
+      "step": 1853
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5344040989875793,
+      "learning_rate": 3.8133718135405285e-06,
+      "loss": 0.1148,
+      "step": 1854
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.688848078250885,
+      "learning_rate": 3.784018903268588e-06,
+      "loss": 0.2293,
+      "step": 1855
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.42069360613822937,
+      "learning_rate": 3.754774956379614e-06,
+      "loss": 0.1282,
+      "step": 1856
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.594592273235321,
+      "learning_rate": 3.7256400418220262e-06,
+      "loss": 0.219,
+      "step": 1857
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.6048558354377747,
+      "learning_rate": 3.696614228287187e-06,
+      "loss": 0.2097,
+      "step": 1858
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5264736413955688,
+      "learning_rate": 3.6676975842092508e-06,
+      "loss": 0.1709,
+      "step": 1859
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5445122122764587,
+      "learning_rate": 3.6388901777649477e-06,
+      "loss": 0.1846,
+      "step": 1860
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5658544301986694,
+      "learning_rate": 3.6101920768734975e-06,
+      "loss": 0.1747,
+      "step": 1861
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5027133226394653,
+      "learning_rate": 3.581603349196372e-06,
+      "loss": 0.1387,
+      "step": 1862
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5999326705932617,
+      "learning_rate": 3.553124062137203e-06,
+      "loss": 0.2154,
+      "step": 1863
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5310497283935547,
+      "learning_rate": 3.5247542828415747e-06,
+      "loss": 0.1799,
+      "step": 1864
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.48598185181617737,
+      "learning_rate": 3.4964940781969104e-06,
+      "loss": 0.1262,
+      "step": 1865
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5608770847320557,
+      "learning_rate": 3.468343514832251e-06,
+      "loss": 0.1866,
+      "step": 1866
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.6633475422859192,
+      "learning_rate": 3.440302659118172e-06,
+      "loss": 0.2332,
+      "step": 1867
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.4052681624889374,
+      "learning_rate": 3.4123715771665786e-06,
+      "loss": 0.093,
+      "step": 1868
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.6034460663795471,
+      "learning_rate": 3.384550334830555e-06,
+      "loss": 0.181,
+      "step": 1869
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5648466348648071,
+      "learning_rate": 3.356838997704226e-06,
+      "loss": 0.1657,
+      "step": 1870
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5791301727294922,
+      "learning_rate": 3.3292376311225835e-06,
+      "loss": 0.1416,
+      "step": 1871
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.4571012258529663,
+      "learning_rate": 3.3017463001613626e-06,
+      "loss": 0.1172,
+      "step": 1872
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.6213254928588867,
+      "learning_rate": 3.274365069636831e-06,
+      "loss": 0.1651,
+      "step": 1873
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5217423439025879,
+      "learning_rate": 3.2470940041057106e-06,
+      "loss": 0.1503,
+      "step": 1874
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.4638414978981018,
+      "learning_rate": 3.2199331678649803e-06,
+      "loss": 0.1737,
+      "step": 1875
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5715259313583374,
+      "learning_rate": 3.1928826249516987e-06,
+      "loss": 0.177,
+      "step": 1876
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.4767167270183563,
+      "learning_rate": 3.165942439142927e-06,
+      "loss": 0.1289,
+      "step": 1877
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5609049201011658,
+      "learning_rate": 3.139112673955513e-06,
+      "loss": 0.174,
+      "step": 1878
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5182044506072998,
+      "learning_rate": 3.112393392645985e-06,
+      "loss": 0.164,
+      "step": 1879
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5244629979133606,
+      "learning_rate": 3.08578465821035e-06,
+      "loss": 0.1569,
+      "step": 1880
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.6161662340164185,
+      "learning_rate": 3.0592865333840204e-06,
+      "loss": 0.1648,
+      "step": 1881
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5368185043334961,
+      "learning_rate": 3.0328990806415934e-06,
+      "loss": 0.1704,
+      "step": 1882
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5800496339797974,
+      "learning_rate": 3.006622362196748e-06,
+      "loss": 0.1632,
+      "step": 1883
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5610716938972473,
+      "learning_rate": 2.9804564400020994e-06,
+      "loss": 0.1452,
+      "step": 1884
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.6072638034820557,
+      "learning_rate": 2.954401375748994e-06,
+      "loss": 0.1961,
+      "step": 1885
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.49570244550704956,
+      "learning_rate": 2.9284572308674596e-06,
+      "loss": 0.1484,
+      "step": 1886
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.7767899036407471,
+      "learning_rate": 2.9026240665259717e-06,
+      "loss": 0.302,
+      "step": 1887
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5587995052337646,
+      "learning_rate": 2.8769019436313715e-06,
+      "loss": 0.1603,
+      "step": 1888
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5553686022758484,
+      "learning_rate": 2.851290922828681e-06,
+      "loss": 0.1457,
+      "step": 1889
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.7426819801330566,
+      "learning_rate": 2.825791064500993e-06,
+      "loss": 0.183,
+      "step": 1890
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.4891470968723297,
+      "learning_rate": 2.800402428769294e-06,
+      "loss": 0.1778,
+      "step": 1891
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5302889943122864,
+      "learning_rate": 2.7751250754923573e-06,
+      "loss": 0.1711,
+      "step": 1892
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.6757040619850159,
+      "learning_rate": 2.7499590642665774e-06,
+      "loss": 0.2504,
+      "step": 1893
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5467231273651123,
+      "learning_rate": 2.7249044544258363e-06,
+      "loss": 0.1922,
+      "step": 1894
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5042440295219421,
+      "learning_rate": 2.699961305041382e-06,
+      "loss": 0.1579,
+      "step": 1895
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.7153398990631104,
+      "learning_rate": 2.675129674921639e-06,
+      "loss": 0.1753,
+      "step": 1896
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.568280041217804,
+      "learning_rate": 2.6504096226121376e-06,
+      "loss": 0.2009,
+      "step": 1897
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5615457892417908,
+      "learning_rate": 2.625801206395312e-06,
+      "loss": 0.1682,
+      "step": 1898
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5228293538093567,
+      "learning_rate": 2.601304484290423e-06,
+      "loss": 0.2239,
+      "step": 1899
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.4459281265735626,
+      "learning_rate": 2.576919514053355e-06,
+      "loss": 0.1272,
+      "step": 1900
+    },
+    {
+      "epoch": 1.8,
+      "eval_loss": 0.2826477885246277,
+      "eval_runtime": 195.5624,
+      "eval_samples_per_second": 1.744,
+      "eval_steps_per_second": 0.22,
+      "step": 1900
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.47112783789634705,
+      "learning_rate": 2.5526463531765465e-06,
+      "loss": 0.1284,
+      "step": 1901
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5726578235626221,
+      "learning_rate": 2.528485058888813e-06,
+      "loss": 0.1686,
+      "step": 1902
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5775447487831116,
+      "learning_rate": 2.5044356881552044e-06,
+      "loss": 0.1855,
+      "step": 1903
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.4397730231285095,
+      "learning_rate": 2.48049829767692e-06,
+      "loss": 0.1215,
+      "step": 1904
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.5429428815841675,
+      "learning_rate": 2.456672943891114e-06,
+      "loss": 0.1802,
+      "step": 1905
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.7085356116294861,
+      "learning_rate": 2.4329596829708144e-06,
+      "loss": 0.2373,
+      "step": 1906
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.519445538520813,
+      "learning_rate": 2.4093585708247492e-06,
+      "loss": 0.1262,
+      "step": 1907
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.5588938593864441,
+      "learning_rate": 2.385869663097251e-06,
+      "loss": 0.1553,
+      "step": 1908
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.5229123830795288,
+      "learning_rate": 2.3624930151680878e-06,
+      "loss": 0.1772,
+      "step": 1909
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.4331282675266266,
+      "learning_rate": 2.3392286821523724e-06,
+      "loss": 0.1121,
+      "step": 1910
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.5925610065460205,
+      "learning_rate": 2.316076718900412e-06,
+      "loss": 0.1682,
+      "step": 1911
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.6164054870605469,
+      "learning_rate": 2.2930371799975594e-06,
+      "loss": 0.1895,
+      "step": 1912
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.6649594306945801,
+      "learning_rate": 2.2701101197641237e-06,
+      "loss": 0.2198,
+      "step": 1913
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.4461248219013214,
+      "learning_rate": 2.2472955922552163e-06,
+      "loss": 0.1467,
+      "step": 1914
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.45993733406066895,
+      "learning_rate": 2.2245936512606314e-06,
+      "loss": 0.1192,
+      "step": 1915
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5341062545776367,
+      "learning_rate": 2.202004350304715e-06,
+      "loss": 0.1671,
+      "step": 1916
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5247201323509216,
+      "learning_rate": 2.179527742646248e-06,
+      "loss": 0.1478,
+      "step": 1917
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.4954621493816376,
+      "learning_rate": 2.157163881278312e-06,
+      "loss": 0.1565,
+      "step": 1918
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.3571833372116089,
+      "learning_rate": 2.1349128189281586e-06,
+      "loss": 0.0895,
+      "step": 1919
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.6149205565452576,
+      "learning_rate": 2.112774608057111e-06,
+      "loss": 0.1901,
+      "step": 1920
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5224018096923828,
+      "learning_rate": 2.0907493008604007e-06,
+      "loss": 0.1511,
+      "step": 1921
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.6662312746047974,
+      "learning_rate": 2.068836949267089e-06,
+      "loss": 0.2431,
+      "step": 1922
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.4910092055797577,
+      "learning_rate": 2.047037604939894e-06,
+      "loss": 0.1731,
+      "step": 1923
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.539424479007721,
+      "learning_rate": 2.0253513192751373e-06,
+      "loss": 0.1585,
+      "step": 1924
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.4712909758090973,
+      "learning_rate": 2.003778143402535e-06,
+      "loss": 0.1363,
+      "step": 1925
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5580833554267883,
+      "learning_rate": 1.982318128185151e-06,
+      "loss": 0.1848,
+      "step": 1926
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.563349187374115,
+      "learning_rate": 1.960971324219263e-06,
+      "loss": 0.1639,
+      "step": 1927
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6639775633811951,
+      "learning_rate": 1.9397377818341944e-06,
+      "loss": 0.22,
+      "step": 1928
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5593473315238953,
+      "learning_rate": 1.9186175510922664e-06,
+      "loss": 0.1625,
+      "step": 1929
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.7248660326004028,
+      "learning_rate": 1.8976106817886196e-06,
+      "loss": 0.2382,
+      "step": 1930
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6270869374275208,
+      "learning_rate": 1.876717223451141e-06,
+      "loss": 0.21,
+      "step": 1931
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6246365904808044,
+      "learning_rate": 1.855937225340315e-06,
+      "loss": 0.175,
+      "step": 1932
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6057000756263733,
+      "learning_rate": 1.835270736449135e-06,
+      "loss": 0.179,
+      "step": 1933
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.4754577577114105,
+      "learning_rate": 1.8147178055029579e-06,
+      "loss": 0.1412,
+      "step": 1934
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5642787218093872,
+      "learning_rate": 1.794278480959416e-06,
+      "loss": 0.1969,
+      "step": 1935
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5310706496238708,
+      "learning_rate": 1.7739528110083004e-06,
+      "loss": 0.1768,
+      "step": 1936
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.49864649772644043,
+      "learning_rate": 1.7537408435714053e-06,
+      "loss": 0.1372,
+      "step": 1937
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.6254642009735107,
+      "learning_rate": 1.7336426263024897e-06,
+      "loss": 0.2255,
+      "step": 1938
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.45464232563972473,
+      "learning_rate": 1.7136582065870877e-06,
+      "loss": 0.1322,
+      "step": 1939
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.5122743844985962,
+      "learning_rate": 1.6937876315424705e-06,
+      "loss": 0.1785,
+      "step": 1940
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.42019030451774597,
+      "learning_rate": 1.674030948017463e-06,
+      "loss": 0.1318,
+      "step": 1941
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.625495970249176,
+      "learning_rate": 1.6543882025923886e-06,
+      "loss": 0.1636,
+      "step": 1942
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.5979576110839844,
+      "learning_rate": 1.6348594415789288e-06,
+      "loss": 0.1766,
+      "step": 1943
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.7714709043502808,
+      "learning_rate": 1.6154447110200255e-06,
+      "loss": 0.1556,
+      "step": 1944
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.5994133353233337,
+      "learning_rate": 1.5961440566897912e-06,
+      "loss": 0.1644,
+      "step": 1945
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.4960111081600189,
+      "learning_rate": 1.5769575240933422e-06,
+      "loss": 0.1683,
+      "step": 1946
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.479365736246109,
+      "learning_rate": 1.5578851584667652e-06,
+      "loss": 0.1412,
+      "step": 1947
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5393126010894775,
+      "learning_rate": 1.5389270047769578e-06,
+      "loss": 0.1366,
+      "step": 1948
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.4532238245010376,
+      "learning_rate": 1.5200831077215428e-06,
+      "loss": 0.1171,
+      "step": 1949
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6420217156410217,
+      "learning_rate": 1.5013535117287647e-06,
+      "loss": 0.2263,
+      "step": 1950
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.606155276298523,
+      "learning_rate": 1.4827382609573781e-06,
+      "loss": 0.2119,
+      "step": 1951
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6354537606239319,
+      "learning_rate": 1.4642373992965364e-06,
+      "loss": 0.1951,
+      "step": 1952
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6020841598510742,
+      "learning_rate": 1.4458509703657196e-06,
+      "loss": 0.1693,
+      "step": 1953
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5179505944252014,
+      "learning_rate": 1.4275790175145908e-06,
+      "loss": 0.1599,
+      "step": 1954
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5395336151123047,
+      "learning_rate": 1.4094215838229176e-06,
+      "loss": 0.1815,
+      "step": 1955
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5530725121498108,
+      "learning_rate": 1.3913787121004719e-06,
+      "loss": 0.196,
+      "step": 1956
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.570218026638031,
+      "learning_rate": 1.3734504448869145e-06,
+      "loss": 0.1526,
+      "step": 1957
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5520387291908264,
+      "learning_rate": 1.3556368244517114e-06,
+      "loss": 0.1657,
+      "step": 1958
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6355603933334351,
+      "learning_rate": 1.3379378927940167e-06,
+      "loss": 0.1949,
+      "step": 1959
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.580990731716156,
+      "learning_rate": 1.3203536916425841e-06,
+      "loss": 0.1867,
+      "step": 1960
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5239478349685669,
+      "learning_rate": 1.3028842624556893e-06,
+      "loss": 0.1686,
+      "step": 1961
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.7929992079734802,
+      "learning_rate": 1.2855296464209688e-06,
+      "loss": 0.2535,
+      "step": 1962
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5575469732284546,
+      "learning_rate": 1.268289884455409e-06,
+      "loss": 0.1604,
+      "step": 1963
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6319878697395325,
+      "learning_rate": 1.2511650172051637e-06,
+      "loss": 0.2069,
+      "step": 1964
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.4371590316295624,
+      "learning_rate": 1.2341550850455352e-06,
+      "loss": 0.1169,
+      "step": 1965
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6503375172615051,
+      "learning_rate": 1.217260128080816e-06,
+      "loss": 0.1919,
+      "step": 1966
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5106748342514038,
+      "learning_rate": 1.2004801861442371e-06,
+      "loss": 0.1383,
+      "step": 1967
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.663134753704071,
+      "learning_rate": 1.1838152987978578e-06,
+      "loss": 0.2009,
+      "step": 1968
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5583426356315613,
+      "learning_rate": 1.1672655053324655e-06,
+      "loss": 0.1653,
+      "step": 1969
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.6124983429908752,
+      "learning_rate": 1.1508308447674975e-06,
+      "loss": 0.1924,
+      "step": 1970
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5914346575737,
+      "learning_rate": 1.1345113558509423e-06,
+      "loss": 0.1994,
+      "step": 1971
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.637253999710083,
+      "learning_rate": 1.1183070770592441e-06,
+      "loss": 0.1988,
+      "step": 1972
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5336123704910278,
+      "learning_rate": 1.1022180465972199e-06,
+      "loss": 0.1622,
+      "step": 1973
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.4567291736602783,
+      "learning_rate": 1.0862443023979651e-06,
+      "loss": 0.1448,
+      "step": 1974
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5192530751228333,
+      "learning_rate": 1.0703858821227541e-06,
+      "loss": 0.1672,
+      "step": 1975
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5903334021568298,
+      "learning_rate": 1.0546428231609895e-06,
+      "loss": 0.1812,
+      "step": 1976
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5002025961875916,
+      "learning_rate": 1.0390151626300526e-06,
+      "loss": 0.1659,
+      "step": 1977
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.6002932190895081,
+      "learning_rate": 1.0235029373752758e-06,
+      "loss": 0.199,
+      "step": 1978
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5473059415817261,
+      "learning_rate": 1.0081061839698258e-06,
+      "loss": 0.1525,
+      "step": 1979
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.46827834844589233,
+      "learning_rate": 9.928249387145982e-07,
+      "loss": 0.1449,
+      "step": 1980
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.6070404648780823,
+      "learning_rate": 9.776592376381955e-07,
+      "loss": 0.2005,
+      "step": 1981
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.42876869440078735,
+      "learning_rate": 9.6260911649676e-07,
+      "loss": 0.1229,
+      "step": 1982
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5290343761444092,
+      "learning_rate": 9.476746107739576e-07,
+      "loss": 0.1318,
+      "step": 1983
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.6454979777336121,
+      "learning_rate": 9.32855755680867e-07,
+      "loss": 0.2159,
+      "step": 1984
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.4990655779838562,
+      "learning_rate": 9.181525861558849e-07,
+      "loss": 0.1728,
+      "step": 1985
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.4970451891422272,
+      "learning_rate": 9.035651368646648e-07,
+      "loss": 0.1761,
+      "step": 1986
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.7014318704605103,
+      "learning_rate": 8.89093442200023e-07,
+      "loss": 0.2338,
+      "step": 1987
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5497717261314392,
+      "learning_rate": 8.747375362818666e-07,
+      "loss": 0.1246,
+      "step": 1988
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.6076444387435913,
+      "learning_rate": 8.604974529571042e-07,
+      "loss": 0.1814,
+      "step": 1989
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.6378231048583984,
+      "learning_rate": 8.463732257995571e-07,
+      "loss": 0.2145,
+      "step": 1990
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.7256751656532288,
+      "learning_rate": 8.323648881099211e-07,
+      "loss": 0.2424,
+      "step": 1991
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.48292362689971924,
+      "learning_rate": 8.184724729156379e-07,
+      "loss": 0.1352,
+      "step": 1992
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.6443254947662354,
+      "learning_rate": 8.046960129708348e-07,
+      "loss": 0.2011,
+      "step": 1993
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.6074917316436768,
+      "learning_rate": 7.910355407562742e-07,
+      "loss": 0.1875,
+      "step": 1994
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.5525424480438232,
+      "learning_rate": 7.774910884792319e-07,
+      "loss": 0.1975,
+      "step": 1995
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4899372160434723,
+      "learning_rate": 7.640626880734581e-07,
+      "loss": 0.1272,
+      "step": 1996
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4794888496398926,
+      "learning_rate": 7.507503711990771e-07,
+      "loss": 0.1391,
+      "step": 1997
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.555847704410553,
+      "learning_rate": 7.375541692425326e-07,
+      "loss": 0.1656,
+      "step": 1998
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.6117456555366516,
+      "learning_rate": 7.244741133164978e-07,
+      "loss": 0.179,
+      "step": 1999
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.6276198625564575,
+      "learning_rate": 7.115102342598101e-07,
+      "loss": 0.2144,
+      "step": 2000
+    },
+    {
+      "epoch": 1.9,
+      "eval_loss": 0.28352752327919006,
+      "eval_runtime": 196.0917,
+      "eval_samples_per_second": 1.739,
+      "eval_steps_per_second": 0.219,
+      "step": 2000
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.4649946689605713,
+      "learning_rate": 6.986625626373977e-07,
+      "loss": 0.1455,
+      "step": 2001
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5064752101898193,
+      "learning_rate": 6.859311287402081e-07,
+      "loss": 0.1342,
+      "step": 2002
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.524705708026886,
+      "learning_rate": 6.733159625851304e-07,
+      "loss": 0.1607,
+      "step": 2003
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5456307530403137,
+      "learning_rate": 6.608170939149283e-07,
+      "loss": 0.1677,
+      "step": 2004
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5111620426177979,
+      "learning_rate": 6.484345521981849e-07,
+      "loss": 0.139,
+      "step": 2005
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5046352744102478,
+      "learning_rate": 6.361683666291973e-07,
+      "loss": 0.1149,
+      "step": 2006
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.49723055958747864,
+      "learning_rate": 6.24018566127954e-07,
+      "loss": 0.1439,
+      "step": 2007
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.46500396728515625,
+      "learning_rate": 6.119851793400189e-07,
+      "loss": 0.1296,
+      "step": 2008
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.6705750823020935,
+      "learning_rate": 6.000682346365083e-07,
+      "loss": 0.1766,
+      "step": 2009
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.553641140460968,
+      "learning_rate": 5.882677601139919e-07,
+      "loss": 0.2046,
+      "step": 2010
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5599265694618225,
+      "learning_rate": 5.765837835944309e-07,
+      "loss": 0.1592,
+      "step": 2011
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6053599119186401,
+      "learning_rate": 5.650163326251345e-07,
+      "loss": 0.1944,
+      "step": 2012
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.69095379114151,
+      "learning_rate": 5.535654344786756e-07,
+      "loss": 0.2488,
+      "step": 2013
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5095921158790588,
+      "learning_rate": 5.422311161528193e-07,
+      "loss": 0.1571,
+      "step": 2014
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.7098842263221741,
+      "learning_rate": 5.310134043704896e-07,
+      "loss": 0.2023,
+      "step": 2015
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5252493619918823,
+      "learning_rate": 5.199123255796634e-07,
+      "loss": 0.1192,
+      "step": 2016
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6396733522415161,
+      "learning_rate": 5.089279059533658e-07,
+      "loss": 0.2144,
+      "step": 2017
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5656467080116272,
+      "learning_rate": 4.980601713895305e-07,
+      "loss": 0.1991,
+      "step": 2018
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5127898454666138,
+      "learning_rate": 4.873091475110281e-07,
+      "loss": 0.1365,
+      "step": 2019
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.47322818636894226,
+      "learning_rate": 4.7667485966552685e-07,
+      "loss": 0.163,
+      "step": 2020
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5779179334640503,
+      "learning_rate": 4.66157332925482e-07,
+      "loss": 0.1817,
+      "step": 2021
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.47470223903656006,
+      "learning_rate": 4.55756592088058e-07,
+      "loss": 0.1364,
+      "step": 2022
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.47271987795829773,
+      "learning_rate": 4.454726616750726e-07,
+      "loss": 0.154,
+      "step": 2023
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.594332754611969,
+      "learning_rate": 4.353055659329419e-07,
+      "loss": 0.1956,
+      "step": 2024
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6354993581771851,
+      "learning_rate": 4.2525532883261886e-07,
+      "loss": 0.1842,
+      "step": 2025
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6601294875144958,
+      "learning_rate": 4.153219740695435e-07,
+      "loss": 0.1668,
+      "step": 2026
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5650955438613892,
+      "learning_rate": 4.0550552506357644e-07,
+      "loss": 0.1556,
+      "step": 2027
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5558925867080688,
+      "learning_rate": 3.958060049589485e-07,
+      "loss": 0.1776,
+      "step": 2028
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.538788914680481,
+      "learning_rate": 3.862234366242168e-07,
+      "loss": 0.1648,
+      "step": 2029
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6010963916778564,
+      "learning_rate": 3.767578426521923e-07,
+      "loss": 0.1922,
+      "step": 2030
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6341791749000549,
+      "learning_rate": 3.674092453598954e-07,
+      "loss": 0.174,
+      "step": 2031
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.4688526391983032,
+      "learning_rate": 3.5817766678850615e-07,
+      "loss": 0.119,
+      "step": 2032
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5660356283187866,
+      "learning_rate": 3.490631287033197e-07,
+      "loss": 0.1786,
+      "step": 2033
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5568372011184692,
+      "learning_rate": 3.4006565259366295e-07,
+      "loss": 0.18,
+      "step": 2034
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.546704113483429,
+      "learning_rate": 3.3118525967289483e-07,
+      "loss": 0.1888,
+      "step": 2035
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5646169185638428,
+      "learning_rate": 3.224219708782894e-07,
+      "loss": 0.1435,
+      "step": 2036
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5710260272026062,
+      "learning_rate": 3.1377580687106944e-07,
+      "loss": 0.1943,
+      "step": 2037
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.7230275869369507,
+      "learning_rate": 3.052467880362675e-07,
+      "loss": 0.2402,
+      "step": 2038
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.645719587802887,
+      "learning_rate": 2.968349344827592e-07,
+      "loss": 0.2287,
+      "step": 2039
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5272510647773743,
+      "learning_rate": 2.88540266043158e-07,
+      "loss": 0.1205,
+      "step": 2040
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.8667675256729126,
+      "learning_rate": 2.8036280227379806e-07,
+      "loss": 0.3002,
+      "step": 2041
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.6541191935539246,
+      "learning_rate": 2.723025624546849e-07,
+      "loss": 0.2162,
+      "step": 2042
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.6151245832443237,
+      "learning_rate": 2.6435956558943374e-07,
+      "loss": 0.1839,
+      "step": 2043
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.629209578037262,
+      "learning_rate": 2.5653383040524227e-07,
+      "loss": 0.1872,
+      "step": 2044
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.5113064050674438,
+      "learning_rate": 2.488253753528458e-07,
+      "loss": 0.1404,
+      "step": 2045
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3060895800590515,
+      "learning_rate": 2.412342186064565e-07,
+      "loss": 0.0703,
+      "step": 2046
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.5913242101669312,
+      "learning_rate": 2.3376037806374097e-07,
+      "loss": 0.1874,
+      "step": 2047
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.6508937478065491,
+      "learning_rate": 2.2640387134577058e-07,
+      "loss": 0.1444,
+      "step": 2048
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.48249170184135437,
+      "learning_rate": 2.1916471579697117e-07,
+      "loss": 0.1365,
+      "step": 2049
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.45732811093330383,
+      "learning_rate": 2.1204292848509555e-07,
+      "loss": 0.1235,
+      "step": 2050
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.6454253196716309,
+      "learning_rate": 2.0503852620117892e-07,
+      "loss": 0.2309,
+      "step": 2051
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.6204021573066711,
+      "learning_rate": 1.98151525459489e-07,
+      "loss": 0.1657,
+      "step": 2052
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5247546434402466,
+      "learning_rate": 1.9138194249750386e-07,
+      "loss": 0.1532,
+      "step": 2053
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.45594534277915955,
+      "learning_rate": 1.8472979327587292e-07,
+      "loss": 0.1251,
+      "step": 2054
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.48648107051849365,
+      "learning_rate": 1.781950934783505e-07,
+      "loss": 0.1426,
+      "step": 2055
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.48606571555137634,
+      "learning_rate": 1.717778585118013e-07,
+      "loss": 0.1206,
+      "step": 2056
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4808293581008911,
+      "learning_rate": 1.6547810350613368e-07,
+      "loss": 0.1481,
+      "step": 2057
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.6266642212867737,
+      "learning_rate": 1.5929584331427216e-07,
+      "loss": 0.1601,
+      "step": 2058
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5545297265052795,
+      "learning_rate": 1.5323109251212943e-07,
+      "loss": 0.1392,
+      "step": 2059
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.48889464139938354,
+      "learning_rate": 1.4728386539856754e-07,
+      "loss": 0.1594,
+      "step": 2060
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5769791007041931,
+      "learning_rate": 1.4145417599534804e-07,
+      "loss": 0.1625,
+      "step": 2061
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4970422387123108,
+      "learning_rate": 1.3574203804713747e-07,
+      "loss": 0.1458,
+      "step": 2062
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5654568672180176,
+      "learning_rate": 1.3014746502142962e-07,
+      "loss": 0.1546,
+      "step": 2063
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.0464012622833252,
+      "learning_rate": 1.246704701085566e-07,
+      "loss": 0.2271,
+      "step": 2064
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.48310214281082153,
+      "learning_rate": 1.1931106622161124e-07,
+      "loss": 0.1281,
+      "step": 2065
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.501817524433136,
+      "learning_rate": 1.1406926599646373e-07,
+      "loss": 0.1309,
+      "step": 2066
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.527482807636261,
+      "learning_rate": 1.0894508179170038e-07,
+      "loss": 0.159,
+      "step": 2067
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.4563215374946594,
+      "learning_rate": 1.0393852568860719e-07,
+      "loss": 0.1391,
+      "step": 2068
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.6645215749740601,
+      "learning_rate": 9.904960949114194e-08,
+      "loss": 0.2544,
+      "step": 2069
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5916361212730408,
+      "learning_rate": 9.427834472588992e-08,
+      "loss": 0.1683,
+      "step": 2070
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.7358233332633972,
+      "learning_rate": 8.962474264206377e-08,
+      "loss": 0.2828,
+      "step": 2071
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.646080732345581,
+      "learning_rate": 8.508881421145365e-08,
+      "loss": 0.1789,
+      "step": 2072
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.6179900169372559,
+      "learning_rate": 8.067057012842161e-08,
+      "loss": 0.1971,
+      "step": 2073
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5118117928504944,
+      "learning_rate": 7.637002080985168e-08,
+      "loss": 0.1604,
+      "step": 2074
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5297279953956604,
+      "learning_rate": 7.218717639514983e-08,
+      "loss": 0.1363,
+      "step": 2075
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6859990954399109,
+      "learning_rate": 6.812204674619959e-08,
+      "loss": 0.2071,
+      "step": 2076
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6707828640937805,
+      "learning_rate": 6.417464144736208e-08,
+      "loss": 0.2514,
+      "step": 2077
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.44010797142982483,
+      "learning_rate": 6.034496980542037e-08,
+      "loss": 0.0956,
+      "step": 2078
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5418484807014465,
+      "learning_rate": 5.663304084960186e-08,
+      "loss": 0.1174,
+      "step": 2079
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.4791770577430725,
+      "learning_rate": 5.303886333151153e-08,
+      "loss": 0.1293,
+      "step": 2080
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5524902939796448,
+      "learning_rate": 4.9562445725132026e-08,
+      "loss": 0.1632,
+      "step": 2081
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.465320348739624,
+      "learning_rate": 4.620379622682358e-08,
+      "loss": 0.1463,
+      "step": 2082
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.47619831562042236,
+      "learning_rate": 4.296292275526859e-08,
+      "loss": 0.1232,
+      "step": 2083
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.7487711310386658,
+      "learning_rate": 3.9839832951465986e-08,
+      "loss": 0.2345,
+      "step": 2084
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5918759703636169,
+      "learning_rate": 3.6834534178725734e-08,
+      "loss": 0.2052,
+      "step": 2085
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.6639288067817688,
+      "learning_rate": 3.394703352263551e-08,
+      "loss": 0.1813,
+      "step": 2086
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5806013345718384,
+      "learning_rate": 3.117733779105514e-08,
+      "loss": 0.1905,
+      "step": 2087
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.43504631519317627,
+      "learning_rate": 2.8525453514099965e-08,
+      "loss": 0.1577,
+      "step": 2088
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5881670117378235,
+      "learning_rate": 2.599138694410752e-08,
+      "loss": 0.1776,
+      "step": 2089
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.6287667751312256,
+      "learning_rate": 2.3575144055643094e-08,
+      "loss": 0.1935,
+      "step": 2090
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5495849847793579,
+      "learning_rate": 2.127673054548862e-08,
+      "loss": 0.1727,
+      "step": 2091
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.585745632648468,
+      "learning_rate": 1.9096151832609375e-08,
+      "loss": 0.1814,
+      "step": 2092
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5768430233001709,
+      "learning_rate": 1.703341305815398e-08,
+      "loss": 0.196,
+      "step": 2093
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.58588045835495,
+      "learning_rate": 1.5088519085437734e-08,
+      "loss": 0.209,
+      "step": 2094
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5062113404273987,
+      "learning_rate": 1.3261474499937087e-08,
+      "loss": 0.126,
+      "step": 2095
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5870720148086548,
+      "learning_rate": 1.1552283609272962e-08,
+      "loss": 0.1645,
+      "step": 2096
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.7464112639427185,
+      "learning_rate": 9.96095044320522e-09,
+      "loss": 0.2385,
+      "step": 2097
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.4561821520328522,
+      "learning_rate": 8.487478753615997e-09,
+      "loss": 0.1138,
+      "step": 2098
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5455138683319092,
+      "learning_rate": 7.13187201450971e-09,
+      "loss": 0.1516,
+      "step": 2099
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5966159105300903,
+      "learning_rate": 5.894133422001957e-09,
+      "loss": 0.1919,
+      "step": 2100
+    },
+    {
+      "epoch": 1.99,
+      "eval_loss": 0.28349846601486206,
+      "eval_runtime": 194.5508,
+      "eval_samples_per_second": 1.753,
+      "eval_steps_per_second": 0.221,
+      "step": 2100
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.7391050457954407,
+      "learning_rate": 4.774265894302854e-09,
+      "loss": 0.2042,
+      "step": 2101
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.4726475179195404,
+      "learning_rate": 3.772272071722593e-09,
+      "loss": 0.1165,
+      "step": 2102
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.560475766658783,
+      "learning_rate": 2.888154316671443e-09,
+      "loss": 0.156,
+      "step": 2103
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5133955478668213,
+      "learning_rate": 2.1219147136264382e-09,
+      "loss": 0.113,
+      "step": 2104
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5839717388153076,
+      "learning_rate": 1.473555069148036e-09,
+      "loss": 0.1849,
+      "step": 2105
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.7022972106933594,
+      "learning_rate": 9.43076911874563e-10,
+      "loss": 0.2225,
+      "step": 2106
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5667376518249512,
+      "learning_rate": 5.304814925111146e-10,
+      "loss": 0.1802,
+      "step": 2107
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5542271733283997,
+      "learning_rate": 2.3576978384065584e-10,
+      "loss": 0.1704,
+      "step": 2108
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5787141919136047,
+      "learning_rate": 5.894248070181796e-11,
+      "loss": 0.1554,
+      "step": 2109
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.568240225315094,
+      "learning_rate": 0.0,
+      "loss": 0.1884,
+      "step": 2110
+    },
+    {
+      "epoch": 2.0,
+      "step": 2110,
+      "total_flos": 2.3203939172220928e+17,
+      "train_loss": 0.24637244094639876,
+      "train_runtime": 113650.884,
+      "train_samples_per_second": 0.594,
+      "train_steps_per_second": 0.019
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2110,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 100,
+  "total_flos": 2.3203939172220928e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}