diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/README.md b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..d4576fe074287232d3836bf69c21d3f2593290d9
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/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-SI-nextqa_shuffled-baseline_lora_mlp/adapter_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..6525dbd33aec747ed5887c28753f3457677d46b8
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/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-si",
+  "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",
+    "gate_proj",
+    "k_proj",
+    "down_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}
\ No newline at end of file
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/adapter_model.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..008415eecaa4fae968d520668761c1fda58e12b8
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ac9ca565ca2ade557fc561b6b90c95e6f09a458ad7d19ec4aecbf3e54832737
+size 692127130
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..c9865019d718e38a92edb33ebb0d1aabd0d6a120
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/config.json
@@ -0,0 +1,59 @@
+{
+  "_name_or_path": "/data/wiedmann/hub/models--lmms-lab--llava-onevision-qwen2-7b-si",
+  "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": null,
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_tunable_parts": "mm_mlp_adapter",
+  "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-SI-nextqa_shuffled-baseline_lora_mlp/generation_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..19a297221acb87418d4388a3decef2282c6d7316
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/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-SI-nextqa_shuffled-baseline_lora_mlp/non_lora_trainables.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..75d8bcd686646adc9f047cf719dd1edba5c86c56
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:414ae8b21d12eeb164ab5d091d6835a24cc2f9317ed362396a627f884a3377bf
+size 33964208
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/trainer_state.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..1303918eb11b84f4c3806d85a0147a291c800e7b
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-SI-nextqa_shuffled-baseline_lora_mlp/trainer_state.json
@@ -0,0 +1,7495 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9996446760630108,
+  "eval_steps": 100,
+  "global_step": 1055,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.573533058166504,
+      "learning_rate": 3.125e-07,
+      "loss": 0.9699,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.963333129882812,
+      "learning_rate": 6.25e-07,
+      "loss": 0.8536,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.163477420806885,
+      "learning_rate": 9.375000000000001e-07,
+      "loss": 1.1326,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.327515125274658,
+      "learning_rate": 1.25e-06,
+      "loss": 1.0532,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 16.021699905395508,
+      "learning_rate": 1.5625e-06,
+      "loss": 0.9634,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.020225524902344,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 1.0092,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.418478012084961,
+      "learning_rate": 2.1875000000000002e-06,
+      "loss": 1.007,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.422685146331787,
+      "learning_rate": 2.5e-06,
+      "loss": 0.9613,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.807549953460693,
+      "learning_rate": 2.8125e-06,
+      "loss": 0.8425,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.5089192390441895,
+      "learning_rate": 3.125e-06,
+      "loss": 0.8219,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.570983409881592,
+      "learning_rate": 3.4375e-06,
+      "loss": 0.8622,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.30479621887207,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.7645,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.199153423309326,
+      "learning_rate": 4.0625000000000005e-06,
+      "loss": 0.7611,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.663885593414307,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 0.7056,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.004486083984375,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 0.8725,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.381334066390991,
+      "learning_rate": 5e-06,
+      "loss": 0.65,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.994814395904541,
+      "learning_rate": 5.3125e-06,
+      "loss": 0.757,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5651938915252686,
+      "learning_rate": 5.625e-06,
+      "loss": 0.5595,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.94376802444458,
+      "learning_rate": 5.9375e-06,
+      "loss": 0.5499,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7750961780548096,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4746,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8759979009628296,
+      "learning_rate": 6.5625e-06,
+      "loss": 0.5768,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.339124321937561,
+      "learning_rate": 6.875e-06,
+      "loss": 0.2824,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5315219163894653,
+      "learning_rate": 7.1875e-06,
+      "loss": 0.4034,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9613999128341675,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.6267,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.240362524986267,
+      "learning_rate": 7.8125e-06,
+      "loss": 0.3157,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9875244498252869,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 0.3633,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.034967064857483,
+      "learning_rate": 8.4375e-06,
+      "loss": 0.3584,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9101635217666626,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.4238,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7796413898468018,
+      "learning_rate": 9.0625e-06,
+      "loss": 0.3641,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.849822461605072,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.3487,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6829776763916016,
+      "learning_rate": 9.6875e-06,
+      "loss": 0.3434,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8473533391952515,
+      "learning_rate": 1e-05,
+      "loss": 0.337,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8275583386421204,
+      "learning_rate": 9.999976423021617e-06,
+      "loss": 0.3868,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7024834752082825,
+      "learning_rate": 9.999905692308813e-06,
+      "loss": 0.3611,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8121680617332458,
+      "learning_rate": 9.999787808528639e-06,
+      "loss": 0.4716,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5949275493621826,
+      "learning_rate": 9.999622772792829e-06,
+      "loss": 0.2894,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7868286371231079,
+      "learning_rate": 9.999410586657801e-06,
+      "loss": 0.3925,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6609775424003601,
+      "learning_rate": 9.999151252124639e-06,
+      "loss": 0.3399,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5456623435020447,
+      "learning_rate": 9.998844771639073e-06,
+      "loss": 0.2579,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5806560516357422,
+      "learning_rate": 9.998491148091457e-06,
+      "loss": 0.3218,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6813802123069763,
+      "learning_rate": 9.99809038481674e-06,
+      "loss": 0.3782,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7458508610725403,
+      "learning_rate": 9.997642485594436e-06,
+      "loss": 0.3484,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6288291811943054,
+      "learning_rate": 9.99714745464859e-06,
+      "loss": 0.2999,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6087560057640076,
+      "learning_rate": 9.996605296647737e-06,
+      "loss": 0.2934,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6758581399917603,
+      "learning_rate": 9.996016016704854e-06,
+      "loss": 0.4618,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5289475321769714,
+      "learning_rate": 9.995379620377319e-06,
+      "loss": 0.2343,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.540494441986084,
+      "learning_rate": 9.99469611366685e-06,
+      "loss": 0.2364,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6204575300216675,
+      "learning_rate": 9.993965503019457e-06,
+      "loss": 0.3579,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6385788917541504,
+      "learning_rate": 9.993187795325381e-06,
+      "loss": 0.2899,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5948849320411682,
+      "learning_rate": 9.992362997919016e-06,
+      "loss": 0.2543,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6491946578025818,
+      "learning_rate": 9.991491118578856e-06,
+      "loss": 0.3367,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6449819803237915,
+      "learning_rate": 9.990572165527413e-06,
+      "loss": 0.3244,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7293139100074768,
+      "learning_rate": 9.98960614743114e-06,
+      "loss": 0.3801,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6604413390159607,
+      "learning_rate": 9.988593073400354e-06,
+      "loss": 0.3575,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6392818689346313,
+      "learning_rate": 9.987532952989145e-06,
+      "loss": 0.3391,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6453878283500671,
+      "learning_rate": 9.986425796195287e-06,
+      "loss": 0.2703,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6270106434822083,
+      "learning_rate": 9.985271613460144e-06,
+      "loss": 0.3559,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6105512380599976,
+      "learning_rate": 9.984070415668574e-06,
+      "loss": 0.1861,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6660328507423401,
+      "learning_rate": 9.98282221414882e-06,
+      "loss": 0.3553,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6679720878601074,
+      "learning_rate": 9.981527020672413e-06,
+      "loss": 0.2726,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6528822779655457,
+      "learning_rate": 9.980184847454052e-06,
+      "loss": 0.3589,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8249501585960388,
+      "learning_rate": 9.978795707151492e-06,
+      "loss": 0.3055,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6238577365875244,
+      "learning_rate": 9.977359612865424e-06,
+      "loss": 0.2546,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7078774571418762,
+      "learning_rate": 9.975876578139355e-06,
+      "loss": 0.2807,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6765748858451843,
+      "learning_rate": 9.974346616959476e-06,
+      "loss": 0.4132,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6024010181427002,
+      "learning_rate": 9.972769743754532e-06,
+      "loss": 0.2856,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.510524332523346,
+      "learning_rate": 9.971145973395685e-06,
+      "loss": 0.1758,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.570541262626648,
+      "learning_rate": 9.969475321196374e-06,
+      "loss": 0.1721,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5491730570793152,
+      "learning_rate": 9.967757802912172e-06,
+      "loss": 0.2676,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5094174742698669,
+      "learning_rate": 9.965993434740634e-06,
+      "loss": 0.1943,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5966671705245972,
+      "learning_rate": 9.96418223332115e-06,
+      "loss": 0.2785,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5527228116989136,
+      "learning_rate": 9.962324215734782e-06,
+      "loss": 0.1879,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.627673864364624,
+      "learning_rate": 9.960419399504107e-06,
+      "loss": 0.3464,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6054822206497192,
+      "learning_rate": 9.958467802593046e-06,
+      "loss": 0.222,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5865452289581299,
+      "learning_rate": 9.956469443406707e-06,
+      "loss": 0.2541,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5308388471603394,
+      "learning_rate": 9.954424340791195e-06,
+      "loss": 0.2497,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7129713296890259,
+      "learning_rate": 9.952332514033449e-06,
+      "loss": 0.2928,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6783535480499268,
+      "learning_rate": 9.950193982861048e-06,
+      "loss": 0.2822,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6808055639266968,
+      "learning_rate": 9.948008767442034e-06,
+      "loss": 0.2615,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5542997121810913,
+      "learning_rate": 9.94577688838472e-06,
+      "loss": 0.2276,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6257948875427246,
+      "learning_rate": 9.943498366737487e-06,
+      "loss": 0.299,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6514169573783875,
+      "learning_rate": 9.941173223988603e-06,
+      "loss": 0.2729,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6393204927444458,
+      "learning_rate": 9.938801482065998e-06,
+      "loss": 0.2799,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.664059042930603,
+      "learning_rate": 9.93638316333708e-06,
+      "loss": 0.3243,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6030569672584534,
+      "learning_rate": 9.93391829060851e-06,
+      "loss": 0.2788,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6259936094284058,
+      "learning_rate": 9.93140688712598e-06,
+      "loss": 0.3041,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6863194108009338,
+      "learning_rate": 9.92884897657402e-06,
+      "loss": 0.3346,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6582688689231873,
+      "learning_rate": 9.926244583075748e-06,
+      "loss": 0.2774,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.649311363697052,
+      "learning_rate": 9.923593731192655e-06,
+      "loss": 0.3058,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6434114575386047,
+      "learning_rate": 9.920896445924372e-06,
+      "loss": 0.2798,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5461995601654053,
+      "learning_rate": 9.918152752708437e-06,
+      "loss": 0.1668,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.646434485912323,
+      "learning_rate": 9.915362677420045e-06,
+      "loss": 0.3334,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.728742241859436,
+      "learning_rate": 9.912526246371815e-06,
+      "loss": 0.3727,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6656957864761353,
+      "learning_rate": 9.909643486313533e-06,
+      "loss": 0.3009,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6820513606071472,
+      "learning_rate": 9.906714424431914e-06,
+      "loss": 0.3038,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5719671845436096,
+      "learning_rate": 9.903739088350325e-06,
+      "loss": 0.2252,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.681084156036377,
+      "learning_rate": 9.90071750612854e-06,
+      "loss": 0.2547,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5969502925872803,
+      "learning_rate": 9.897649706262474e-06,
+      "loss": 0.2529,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7019534707069397,
+      "learning_rate": 9.894535717683902e-06,
+      "loss": 0.3078,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6780735850334167,
+      "learning_rate": 9.891375569760205e-06,
+      "loss": 0.3005,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.2637310326099396,
+      "eval_runtime": 327.5464,
+      "eval_samples_per_second": 1.041,
+      "eval_steps_per_second": 0.263,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6152581572532654,
+      "learning_rate": 9.888169292294077e-06,
+      "loss": 0.2279,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6035614013671875,
+      "learning_rate": 9.88491691552325e-06,
+      "loss": 0.2279,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7469601035118103,
+      "learning_rate": 9.881618470120216e-06,
+      "loss": 0.2789,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7219517827033997,
+      "learning_rate": 9.87827398719192e-06,
+      "loss": 0.3135,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6131671667098999,
+      "learning_rate": 9.874883498279485e-06,
+      "loss": 0.1929,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6404304504394531,
+      "learning_rate": 9.871447035357903e-06,
+      "loss": 0.2558,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.622728705406189,
+      "learning_rate": 9.867964630835742e-06,
+      "loss": 0.2261,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5448305010795593,
+      "learning_rate": 9.86443631755483e-06,
+      "loss": 0.2026,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6098849773406982,
+      "learning_rate": 9.860862128789954e-06,
+      "loss": 0.2017,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6181161403656006,
+      "learning_rate": 9.857242098248543e-06,
+      "loss": 0.1947,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7358696460723877,
+      "learning_rate": 9.853576260070348e-06,
+      "loss": 0.2749,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6237342953681946,
+      "learning_rate": 9.849864648827126e-06,
+      "loss": 0.2232,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8139705657958984,
+      "learning_rate": 9.846107299522305e-06,
+      "loss": 0.336,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7234876751899719,
+      "learning_rate": 9.842304247590668e-06,
+      "loss": 0.3555,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6199493408203125,
+      "learning_rate": 9.838455528897998e-06,
+      "loss": 0.2127,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5883256196975708,
+      "learning_rate": 9.834561179740763e-06,
+      "loss": 0.1799,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6576983332633972,
+      "learning_rate": 9.830621236845755e-06,
+      "loss": 0.2601,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6341786980628967,
+      "learning_rate": 9.826635737369752e-06,
+      "loss": 0.2503,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6072645783424377,
+      "learning_rate": 9.82260471889917e-06,
+      "loss": 0.2027,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8956173062324524,
+      "learning_rate": 9.818528219449705e-06,
+      "loss": 0.2389,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.614122211933136,
+      "learning_rate": 9.814406277465969e-06,
+      "loss": 0.197,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7290271520614624,
+      "learning_rate": 9.810238931821139e-06,
+      "loss": 0.2665,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6730625629425049,
+      "learning_rate": 9.806026221816582e-06,
+      "loss": 0.3386,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5942097902297974,
+      "learning_rate": 9.801768187181487e-06,
+      "loss": 0.2707,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6023765802383423,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 0.2218,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7326840162277222,
+      "learning_rate": 9.793116305073292e-06,
+      "loss": 0.2333,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6956497430801392,
+      "learning_rate": 9.788722539194291e-06,
+      "loss": 0.3369,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6568793058395386,
+      "learning_rate": 9.78428361187217e-06,
+      "loss": 0.2236,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7054308652877808,
+      "learning_rate": 9.77979956496953e-06,
+      "loss": 0.2579,
+      "step": 129
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7843027114868164,
+      "learning_rate": 9.775270440774481e-06,
+      "loss": 0.4016,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7426701784133911,
+      "learning_rate": 9.770696282000245e-06,
+      "loss": 0.3249,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7212030291557312,
+      "learning_rate": 9.766077131784764e-06,
+      "loss": 0.3782,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6892423629760742,
+      "learning_rate": 9.761413033690276e-06,
+      "loss": 0.2803,
+      "step": 133
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7250350117683411,
+      "learning_rate": 9.756704031702919e-06,
+      "loss": 0.2911,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5780990123748779,
+      "learning_rate": 9.75195017023231e-06,
+      "loss": 0.2263,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6007035374641418,
+      "learning_rate": 9.74715149411112e-06,
+      "loss": 0.267,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7782276272773743,
+      "learning_rate": 9.742308048594665e-06,
+      "loss": 0.3348,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4514017105102539,
+      "learning_rate": 9.737419879360471e-06,
+      "loss": 0.1335,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.5577970743179321,
+      "learning_rate": 9.732487032507837e-06,
+      "loss": 0.1962,
+      "step": 139
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8191975355148315,
+      "learning_rate": 9.727509554557416e-06,
+      "loss": 0.359,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6398234963417053,
+      "learning_rate": 9.722487492450764e-06,
+      "loss": 0.1976,
+      "step": 141
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7978889346122742,
+      "learning_rate": 9.717420893549902e-06,
+      "loss": 0.3241,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6350768804550171,
+      "learning_rate": 9.712309805636863e-06,
+      "loss": 0.2473,
+      "step": 143
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6668251752853394,
+      "learning_rate": 9.707154276913255e-06,
+      "loss": 0.3091,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.714043915271759,
+      "learning_rate": 9.701954355999791e-06,
+      "loss": 0.2983,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6359931230545044,
+      "learning_rate": 9.696710091935842e-06,
+      "loss": 0.2758,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5917684435844421,
+      "learning_rate": 9.691421534178966e-06,
+      "loss": 0.2057,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6024724841117859,
+      "learning_rate": 9.68608873260445e-06,
+      "loss": 0.2429,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6073502898216248,
+      "learning_rate": 9.680711737504832e-06,
+      "loss": 0.1956,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.696204662322998,
+      "learning_rate": 9.675290599589429e-06,
+      "loss": 0.3212,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6330821514129639,
+      "learning_rate": 9.669825369983865e-06,
+      "loss": 0.1942,
+      "step": 151
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.666441023349762,
+      "learning_rate": 9.664316100229578e-06,
+      "loss": 0.2903,
+      "step": 152
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6631059050559998,
+      "learning_rate": 9.658762842283343e-06,
+      "loss": 0.2909,
+      "step": 153
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5578485727310181,
+      "learning_rate": 9.653165648516777e-06,
+      "loss": 0.2069,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8304031491279602,
+      "learning_rate": 9.647524571715843e-06,
+      "loss": 0.3124,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6066958904266357,
+      "learning_rate": 9.641839665080363e-06,
+      "loss": 0.2585,
+      "step": 156
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6744480133056641,
+      "learning_rate": 9.636110982223505e-06,
+      "loss": 0.232,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7409882545471191,
+      "learning_rate": 9.630338577171282e-06,
+      "loss": 0.2774,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.595603883266449,
+      "learning_rate": 9.624522504362039e-06,
+      "loss": 0.2175,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.647714376449585,
+      "learning_rate": 9.618662818645949e-06,
+      "loss": 0.2964,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5570324063301086,
+      "learning_rate": 9.612759575284483e-06,
+      "loss": 0.1763,
+      "step": 161
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7239076495170593,
+      "learning_rate": 9.606812829949896e-06,
+      "loss": 0.2865,
+      "step": 162
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6651351451873779,
+      "learning_rate": 9.600822638724704e-06,
+      "loss": 0.2219,
+      "step": 163
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6308421492576599,
+      "learning_rate": 9.594789058101154e-06,
+      "loss": 0.2315,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6480094194412231,
+      "learning_rate": 9.588712144980681e-06,
+      "loss": 0.2858,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7034019827842712,
+      "learning_rate": 9.582591956673387e-06,
+      "loss": 0.2438,
+      "step": 166
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6165547370910645,
+      "learning_rate": 9.57642855089749e-06,
+      "loss": 0.2009,
+      "step": 167
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8720855712890625,
+      "learning_rate": 9.570221985778785e-06,
+      "loss": 0.2984,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6759960651397705,
+      "learning_rate": 9.563972319850092e-06,
+      "loss": 0.2318,
+      "step": 169
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.567792534828186,
+      "learning_rate": 9.557679612050708e-06,
+      "loss": 0.181,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7541651129722595,
+      "learning_rate": 9.551343921725844e-06,
+      "loss": 0.3476,
+      "step": 171
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7993423342704773,
+      "learning_rate": 9.544965308626075e-06,
+      "loss": 0.3679,
+      "step": 172
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6110410690307617,
+      "learning_rate": 9.538543832906773e-06,
+      "loss": 0.2089,
+      "step": 173
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.77556973695755,
+      "learning_rate": 9.532079555127532e-06,
+      "loss": 0.285,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7493922114372253,
+      "learning_rate": 9.525572536251608e-06,
+      "loss": 0.3202,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6218746900558472,
+      "learning_rate": 9.519022837645337e-06,
+      "loss": 0.2713,
+      "step": 176
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6792969703674316,
+      "learning_rate": 9.512430521077565e-06,
+      "loss": 0.2558,
+      "step": 177
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6580852270126343,
+      "learning_rate": 9.505795648719049e-06,
+      "loss": 0.2697,
+      "step": 178
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7115439176559448,
+      "learning_rate": 9.499118283141887e-06,
+      "loss": 0.3133,
+      "step": 179
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5898191928863525,
+      "learning_rate": 9.492398487318922e-06,
+      "loss": 0.259,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6370326280593872,
+      "learning_rate": 9.485636324623147e-06,
+      "loss": 0.2238,
+      "step": 181
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7449048161506653,
+      "learning_rate": 9.478831858827105e-06,
+      "loss": 0.2949,
+      "step": 182
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7421145439147949,
+      "learning_rate": 9.471985154102292e-06,
+      "loss": 0.2105,
+      "step": 183
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7774351239204407,
+      "learning_rate": 9.465096275018556e-06,
+      "loss": 0.3231,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6269638538360596,
+      "learning_rate": 9.458165286543477e-06,
+      "loss": 0.2886,
+      "step": 185
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.699456512928009,
+      "learning_rate": 9.451192254041759e-06,
+      "loss": 0.2803,
+      "step": 186
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7909825444221497,
+      "learning_rate": 9.444177243274619e-06,
+      "loss": 0.3949,
+      "step": 187
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7326850891113281,
+      "learning_rate": 9.437120320399158e-06,
+      "loss": 0.2864,
+      "step": 188
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7978874444961548,
+      "learning_rate": 9.430021551967745e-06,
+      "loss": 0.2661,
+      "step": 189
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6970043778419495,
+      "learning_rate": 9.422881004927383e-06,
+      "loss": 0.2223,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.645399272441864,
+      "learning_rate": 9.41569874661908e-06,
+      "loss": 0.2777,
+      "step": 191
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6742087602615356,
+      "learning_rate": 9.408474844777218e-06,
+      "loss": 0.2067,
+      "step": 192
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7820308208465576,
+      "learning_rate": 9.401209367528907e-06,
+      "loss": 0.4041,
+      "step": 193
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.692047119140625,
+      "learning_rate": 9.393902383393347e-06,
+      "loss": 0.2648,
+      "step": 194
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8333490490913391,
+      "learning_rate": 9.386553961281179e-06,
+      "loss": 0.3695,
+      "step": 195
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7040037512779236,
+      "learning_rate": 9.379164170493844e-06,
+      "loss": 0.3136,
+      "step": 196
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8036830425262451,
+      "learning_rate": 9.371733080722911e-06,
+      "loss": 0.2963,
+      "step": 197
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.5481095314025879,
+      "learning_rate": 9.36426076204944e-06,
+      "loss": 0.1438,
+      "step": 198
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7464080452919006,
+      "learning_rate": 9.35674728494331e-06,
+      "loss": 0.2957,
+      "step": 199
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6098821759223938,
+      "learning_rate": 9.349192720262556e-06,
+      "loss": 0.2428,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.2540014684200287,
+      "eval_runtime": 327.1693,
+      "eval_samples_per_second": 1.042,
+      "eval_steps_per_second": 0.263,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6470658779144287,
+      "learning_rate": 9.341597139252698e-06,
+      "loss": 0.269,
+      "step": 201
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6855325102806091,
+      "learning_rate": 9.333960613546079e-06,
+      "loss": 0.2732,
+      "step": 202
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6597403287887573,
+      "learning_rate": 9.326283215161177e-06,
+      "loss": 0.2185,
+      "step": 203
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4843915104866028,
+      "learning_rate": 9.31856501650194e-06,
+      "loss": 0.117,
+      "step": 204
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6510450839996338,
+      "learning_rate": 9.310806090357083e-06,
+      "loss": 0.3108,
+      "step": 205
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5514283180236816,
+      "learning_rate": 9.30300650989942e-06,
+      "loss": 0.1462,
+      "step": 206
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7147676348686218,
+      "learning_rate": 9.295166348685169e-06,
+      "loss": 0.2668,
+      "step": 207
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7435725331306458,
+      "learning_rate": 9.287285680653254e-06,
+      "loss": 0.2922,
+      "step": 208
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6453981399536133,
+      "learning_rate": 9.279364580124615e-06,
+      "loss": 0.2574,
+      "step": 209
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6430134177207947,
+      "learning_rate": 9.271403121801492e-06,
+      "loss": 0.1975,
+      "step": 210
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.89022296667099,
+      "learning_rate": 9.263401380766739e-06,
+      "loss": 0.3669,
+      "step": 211
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.682579755783081,
+      "learning_rate": 9.255359432483106e-06,
+      "loss": 0.2956,
+      "step": 212
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6273362636566162,
+      "learning_rate": 9.247277352792534e-06,
+      "loss": 0.2152,
+      "step": 213
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7749226689338684,
+      "learning_rate": 9.239155217915422e-06,
+      "loss": 0.2967,
+      "step": 214
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5863397717475891,
+      "learning_rate": 9.23099310444994e-06,
+      "loss": 0.1697,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6522620320320129,
+      "learning_rate": 9.222791089371266e-06,
+      "loss": 0.2507,
+      "step": 216
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.64266037940979,
+      "learning_rate": 9.214549250030899e-06,
+      "loss": 0.2172,
+      "step": 217
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.726273238658905,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": 0.3365,
+      "step": 218
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5961911678314209,
+      "learning_rate": 9.197946409848196e-06,
+      "loss": 0.2067,
+      "step": 219
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6414331793785095,
+      "learning_rate": 9.189585565583779e-06,
+      "loss": 0.2531,
+      "step": 220
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5866371989250183,
+      "learning_rate": 9.181185210212034e-06,
+      "loss": 0.2646,
+      "step": 221
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6715410947799683,
+      "learning_rate": 9.172745422954961e-06,
+      "loss": 0.2473,
+      "step": 222
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6602498888969421,
+      "learning_rate": 9.164266283406433e-06,
+      "loss": 0.2778,
+      "step": 223
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7695605158805847,
+      "learning_rate": 9.155747871531444e-06,
+      "loss": 0.259,
+      "step": 224
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7421642541885376,
+      "learning_rate": 9.147190267665361e-06,
+      "loss": 0.2916,
+      "step": 225
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6979179978370667,
+      "learning_rate": 9.13859355251316e-06,
+      "loss": 0.2285,
+      "step": 226
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6683458089828491,
+      "learning_rate": 9.129957807148666e-06,
+      "loss": 0.2239,
+      "step": 227
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8766207098960876,
+      "learning_rate": 9.121283113013794e-06,
+      "loss": 0.3639,
+      "step": 228
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5637826323509216,
+      "learning_rate": 9.112569551917773e-06,
+      "loss": 0.1856,
+      "step": 229
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6588501334190369,
+      "learning_rate": 9.103817206036383e-06,
+      "loss": 0.3037,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8170912861824036,
+      "learning_rate": 9.095026157911166e-06,
+      "loss": 0.2881,
+      "step": 231
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6367344856262207,
+      "learning_rate": 9.086196490448668e-06,
+      "loss": 0.1722,
+      "step": 232
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7489650249481201,
+      "learning_rate": 9.077328286919638e-06,
+      "loss": 0.3079,
+      "step": 233
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7185512185096741,
+      "learning_rate": 9.068421630958254e-06,
+      "loss": 0.2906,
+      "step": 234
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.61532062292099,
+      "learning_rate": 9.059476606561328e-06,
+      "loss": 0.1991,
+      "step": 235
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7276179790496826,
+      "learning_rate": 9.050493298087523e-06,
+      "loss": 0.272,
+      "step": 236
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6644300818443298,
+      "learning_rate": 9.041471790256543e-06,
+      "loss": 0.2747,
+      "step": 237
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.685745120048523,
+      "learning_rate": 9.032412168148345e-06,
+      "loss": 0.2481,
+      "step": 238
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7370786666870117,
+      "learning_rate": 9.023314517202341e-06,
+      "loss": 0.3041,
+      "step": 239
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6082615852355957,
+      "learning_rate": 9.014178923216572e-06,
+      "loss": 0.1761,
+      "step": 240
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.594700038433075,
+      "learning_rate": 9.005005472346923e-06,
+      "loss": 0.2002,
+      "step": 241
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7096708416938782,
+      "learning_rate": 8.995794251106295e-06,
+      "loss": 0.3118,
+      "step": 242
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6614163517951965,
+      "learning_rate": 8.986545346363792e-06,
+      "loss": 0.2483,
+      "step": 243
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6781965494155884,
+      "learning_rate": 8.977258845343904e-06,
+      "loss": 0.271,
+      "step": 244
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7122507095336914,
+      "learning_rate": 8.96793483562569e-06,
+      "loss": 0.254,
+      "step": 245
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6902074217796326,
+      "learning_rate": 8.958573405141932e-06,
+      "loss": 0.2693,
+      "step": 246
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7376943230628967,
+      "learning_rate": 8.949174642178333e-06,
+      "loss": 0.2949,
+      "step": 247
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7361817955970764,
+      "learning_rate": 8.939738635372664e-06,
+      "loss": 0.2617,
+      "step": 248
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8066937327384949,
+      "learning_rate": 8.930265473713939e-06,
+      "loss": 0.3163,
+      "step": 249
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7305431962013245,
+      "learning_rate": 8.920755246541563e-06,
+      "loss": 0.2936,
+      "step": 250
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7218648195266724,
+      "learning_rate": 8.911208043544513e-06,
+      "loss": 0.2888,
+      "step": 251
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6873829960823059,
+      "learning_rate": 8.90162395476046e-06,
+      "loss": 0.2173,
+      "step": 252
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8219372630119324,
+      "learning_rate": 8.89200307057495e-06,
+      "loss": 0.3976,
+      "step": 253
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7167593836784363,
+      "learning_rate": 8.882345481720533e-06,
+      "loss": 0.302,
+      "step": 254
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7573580145835876,
+      "learning_rate": 8.872651279275917e-06,
+      "loss": 0.2741,
+      "step": 255
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7510555982589722,
+      "learning_rate": 8.862920554665098e-06,
+      "loss": 0.2862,
+      "step": 256
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.678774356842041,
+      "learning_rate": 8.853153399656513e-06,
+      "loss": 0.2724,
+      "step": 257
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7355273365974426,
+      "learning_rate": 8.843349906362163e-06,
+      "loss": 0.2834,
+      "step": 258
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8037059903144836,
+      "learning_rate": 8.833510167236747e-06,
+      "loss": 0.335,
+      "step": 259
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.695874810218811,
+      "learning_rate": 8.823634275076792e-06,
+      "loss": 0.2839,
+      "step": 260
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7235242128372192,
+      "learning_rate": 8.813722323019774e-06,
+      "loss": 0.318,
+      "step": 261
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6236990690231323,
+      "learning_rate": 8.803774404543246e-06,
+      "loss": 0.2083,
+      "step": 262
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6476077437400818,
+      "learning_rate": 8.793790613463956e-06,
+      "loss": 0.2701,
+      "step": 263
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4958818554878235,
+      "learning_rate": 8.783771043936949e-06,
+      "loss": 0.1821,
+      "step": 264
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7664861679077148,
+      "learning_rate": 8.773715790454695e-06,
+      "loss": 0.2629,
+      "step": 265
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6198388338088989,
+      "learning_rate": 8.763624947846195e-06,
+      "loss": 0.1988,
+      "step": 266
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7904531359672546,
+      "learning_rate": 8.75349861127608e-06,
+      "loss": 0.307,
+      "step": 267
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7708582282066345,
+      "learning_rate": 8.743336876243712e-06,
+      "loss": 0.2798,
+      "step": 268
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7881653308868408,
+      "learning_rate": 8.733139838582299e-06,
+      "loss": 0.3482,
+      "step": 269
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7888725399971008,
+      "learning_rate": 8.722907594457975e-06,
+      "loss": 0.3321,
+      "step": 270
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7690854668617249,
+      "learning_rate": 8.712640240368899e-06,
+      "loss": 0.3307,
+      "step": 271
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7235127687454224,
+      "learning_rate": 8.702337873144343e-06,
+      "loss": 0.2644,
+      "step": 272
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6718965172767639,
+      "learning_rate": 8.692000589943785e-06,
+      "loss": 0.2135,
+      "step": 273
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7318533062934875,
+      "learning_rate": 8.681628488255986e-06,
+      "loss": 0.311,
+      "step": 274
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6941997408866882,
+      "learning_rate": 8.671221665898074e-06,
+      "loss": 0.2215,
+      "step": 275
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7051814794540405,
+      "learning_rate": 8.660780221014617e-06,
+      "loss": 0.2967,
+      "step": 276
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6508485674858093,
+      "learning_rate": 8.650304252076704e-06,
+      "loss": 0.2548,
+      "step": 277
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5858379006385803,
+      "learning_rate": 8.63979385788101e-06,
+      "loss": 0.202,
+      "step": 278
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7965670228004456,
+      "learning_rate": 8.629249137548873e-06,
+      "loss": 0.3318,
+      "step": 279
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6064817905426025,
+      "learning_rate": 8.61867019052535e-06,
+      "loss": 0.1971,
+      "step": 280
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6326867938041687,
+      "learning_rate": 8.608057116578283e-06,
+      "loss": 0.2048,
+      "step": 281
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7318657040596008,
+      "learning_rate": 8.597410015797358e-06,
+      "loss": 0.3125,
+      "step": 282
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5439643263816833,
+      "learning_rate": 8.586728988593158e-06,
+      "loss": 0.2009,
+      "step": 283
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6920150518417358,
+      "learning_rate": 8.576014135696227e-06,
+      "loss": 0.2299,
+      "step": 284
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6073454022407532,
+      "learning_rate": 8.565265558156101e-06,
+      "loss": 0.2028,
+      "step": 285
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8154141902923584,
+      "learning_rate": 8.554483357340379e-06,
+      "loss": 0.3168,
+      "step": 286
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5363025069236755,
+      "learning_rate": 8.543667634933743e-06,
+      "loss": 0.2054,
+      "step": 287
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6466490626335144,
+      "learning_rate": 8.532818492937014e-06,
+      "loss": 0.2164,
+      "step": 288
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.860065221786499,
+      "learning_rate": 8.521936033666187e-06,
+      "loss": 0.258,
+      "step": 289
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7522770762443542,
+      "learning_rate": 8.511020359751467e-06,
+      "loss": 0.2297,
+      "step": 290
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5030247569084167,
+      "learning_rate": 8.500071574136297e-06,
+      "loss": 0.149,
+      "step": 291
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6503915786743164,
+      "learning_rate": 8.489089780076387e-06,
+      "loss": 0.2363,
+      "step": 292
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7013444900512695,
+      "learning_rate": 8.478075081138746e-06,
+      "loss": 0.2444,
+      "step": 293
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7805238366127014,
+      "learning_rate": 8.467027581200702e-06,
+      "loss": 0.3497,
+      "step": 294
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.72954261302948,
+      "learning_rate": 8.455947384448926e-06,
+      "loss": 0.2815,
+      "step": 295
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6791868209838867,
+      "learning_rate": 8.444834595378434e-06,
+      "loss": 0.1917,
+      "step": 296
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.786886990070343,
+      "learning_rate": 8.433689318791628e-06,
+      "loss": 0.2849,
+      "step": 297
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7145736813545227,
+      "learning_rate": 8.42251165979728e-06,
+      "loss": 0.2544,
+      "step": 298
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6851241588592529,
+      "learning_rate": 8.411301723809563e-06,
+      "loss": 0.2292,
+      "step": 299
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7301487922668457,
+      "learning_rate": 8.400059616547046e-06,
+      "loss": 0.2454,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.24792622029781342,
+      "eval_runtime": 335.2342,
+      "eval_samples_per_second": 1.017,
+      "eval_steps_per_second": 0.257,
+      "step": 300
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6980499029159546,
+      "learning_rate": 8.388785444031695e-06,
+      "loss": 0.2847,
+      "step": 301
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6795322895050049,
+      "learning_rate": 8.37747931258788e-06,
+      "loss": 0.2189,
+      "step": 302
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6904706954956055,
+      "learning_rate": 8.366141328841367e-06,
+      "loss": 0.3023,
+      "step": 303
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6439961791038513,
+      "learning_rate": 8.354771599718313e-06,
+      "loss": 0.1895,
+      "step": 304
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7503674030303955,
+      "learning_rate": 8.34337023244426e-06,
+      "loss": 0.2955,
+      "step": 305
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7018287777900696,
+      "learning_rate": 8.331937334543132e-06,
+      "loss": 0.251,
+      "step": 306
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.821648895740509,
+      "learning_rate": 8.320473013836197e-06,
+      "loss": 0.3457,
+      "step": 307
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7417198419570923,
+      "learning_rate": 8.308977378441072e-06,
+      "loss": 0.3358,
+      "step": 308
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6748823523521423,
+      "learning_rate": 8.297450536770697e-06,
+      "loss": 0.2351,
+      "step": 309
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7352398633956909,
+      "learning_rate": 8.285892597532311e-06,
+      "loss": 0.2514,
+      "step": 310
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7595478892326355,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 0.2877,
+      "step": 311
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6498991847038269,
+      "learning_rate": 8.262683862645804e-06,
+      "loss": 0.2528,
+      "step": 312
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7450823783874512,
+      "learning_rate": 8.25103328587442e-06,
+      "loss": 0.269,
+      "step": 313
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7365174293518066,
+      "learning_rate": 8.239352049286435e-06,
+      "loss": 0.2763,
+      "step": 314
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6681867837905884,
+      "learning_rate": 8.22764026304515e-06,
+      "loss": 0.2462,
+      "step": 315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7543831467628479,
+      "learning_rate": 8.215898037601981e-06,
+      "loss": 0.294,
+      "step": 316
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.784321129322052,
+      "learning_rate": 8.204125483695403e-06,
+      "loss": 0.2842,
+      "step": 317
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6397828459739685,
+      "learning_rate": 8.192322712349917e-06,
+      "loss": 0.2023,
+      "step": 318
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.815455436706543,
+      "learning_rate": 8.180489834875e-06,
+      "loss": 0.2827,
+      "step": 319
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6677094101905823,
+      "learning_rate": 8.168626962864045e-06,
+      "loss": 0.2163,
+      "step": 320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7844486832618713,
+      "learning_rate": 8.156734208193327e-06,
+      "loss": 0.261,
+      "step": 321
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7194960713386536,
+      "learning_rate": 8.144811683020932e-06,
+      "loss": 0.2892,
+      "step": 322
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6769686341285706,
+      "learning_rate": 8.132859499785708e-06,
+      "loss": 0.2651,
+      "step": 323
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7005006074905396,
+      "learning_rate": 8.120877771206201e-06,
+      "loss": 0.268,
+      "step": 324
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8526335954666138,
+      "learning_rate": 8.108866610279595e-06,
+      "loss": 0.3551,
+      "step": 325
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6851252317428589,
+      "learning_rate": 8.09682613028064e-06,
+      "loss": 0.2551,
+      "step": 326
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7191782593727112,
+      "learning_rate": 8.08475644476059e-06,
+      "loss": 0.3117,
+      "step": 327
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7180798053741455,
+      "learning_rate": 8.072657667546136e-06,
+      "loss": 0.3054,
+      "step": 328
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6469159722328186,
+      "learning_rate": 8.060529912738316e-06,
+      "loss": 0.2902,
+      "step": 329
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7266992330551147,
+      "learning_rate": 8.048373294711455e-06,
+      "loss": 0.2849,
+      "step": 330
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6951152086257935,
+      "learning_rate": 8.036187928112087e-06,
+      "loss": 0.2612,
+      "step": 331
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8010523915290833,
+      "learning_rate": 8.023973927857857e-06,
+      "loss": 0.3538,
+      "step": 332
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7954149842262268,
+      "learning_rate": 8.011731409136454e-06,
+      "loss": 0.305,
+      "step": 333
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7332515716552734,
+      "learning_rate": 7.99946048740452e-06,
+      "loss": 0.232,
+      "step": 334
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.799700140953064,
+      "learning_rate": 7.987161278386555e-06,
+      "loss": 0.3147,
+      "step": 335
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6455225348472595,
+      "learning_rate": 7.974833898073832e-06,
+      "loss": 0.2652,
+      "step": 336
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6520224809646606,
+      "learning_rate": 7.962478462723306e-06,
+      "loss": 0.2673,
+      "step": 337
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6774414777755737,
+      "learning_rate": 7.950095088856509e-06,
+      "loss": 0.287,
+      "step": 338
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5675615072250366,
+      "learning_rate": 7.937683893258454e-06,
+      "loss": 0.1705,
+      "step": 339
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7145261764526367,
+      "learning_rate": 7.925244992976538e-06,
+      "loss": 0.2907,
+      "step": 340
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6485288143157959,
+      "learning_rate": 7.912778505319436e-06,
+      "loss": 0.2452,
+      "step": 341
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5317831635475159,
+      "learning_rate": 7.900284547855992e-06,
+      "loss": 0.1613,
+      "step": 342
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7434039115905762,
+      "learning_rate": 7.88776323841411e-06,
+      "loss": 0.262,
+      "step": 343
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9552202224731445,
+      "learning_rate": 7.875214695079647e-06,
+      "loss": 0.3407,
+      "step": 344
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6228988766670227,
+      "learning_rate": 7.862639036195298e-06,
+      "loss": 0.1854,
+      "step": 345
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8214723467826843,
+      "learning_rate": 7.850036380359479e-06,
+      "loss": 0.3061,
+      "step": 346
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7715755105018616,
+      "learning_rate": 7.837406846425205e-06,
+      "loss": 0.3337,
+      "step": 347
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7069976329803467,
+      "learning_rate": 7.824750553498977e-06,
+      "loss": 0.3405,
+      "step": 348
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7876215577125549,
+      "learning_rate": 7.812067620939653e-06,
+      "loss": 0.2986,
+      "step": 349
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6568417549133301,
+      "learning_rate": 7.799358168357323e-06,
+      "loss": 0.2195,
+      "step": 350
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7959166765213013,
+      "learning_rate": 7.786622315612182e-06,
+      "loss": 0.2963,
+      "step": 351
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6602903604507446,
+      "learning_rate": 7.773860182813404e-06,
+      "loss": 0.1999,
+      "step": 352
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7031603455543518,
+      "learning_rate": 7.761071890317994e-06,
+      "loss": 0.2623,
+      "step": 353
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6833639740943909,
+      "learning_rate": 7.748257558729677e-06,
+      "loss": 0.2387,
+      "step": 354
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6679258942604065,
+      "learning_rate": 7.735417308897737e-06,
+      "loss": 0.2951,
+      "step": 355
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6927437782287598,
+      "learning_rate": 7.72255126191589e-06,
+      "loss": 0.2516,
+      "step": 356
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7067151665687561,
+      "learning_rate": 7.709659539121144e-06,
+      "loss": 0.2554,
+      "step": 357
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6713438630104065,
+      "learning_rate": 7.696742262092643e-06,
+      "loss": 0.2542,
+      "step": 358
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5410385131835938,
+      "learning_rate": 7.683799552650534e-06,
+      "loss": 0.1874,
+      "step": 359
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6991607546806335,
+      "learning_rate": 7.670831532854811e-06,
+      "loss": 0.1918,
+      "step": 360
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5746275782585144,
+      "learning_rate": 7.65783832500416e-06,
+      "loss": 0.185,
+      "step": 361
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6405538320541382,
+      "learning_rate": 7.644820051634813e-06,
+      "loss": 0.2428,
+      "step": 362
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7153875827789307,
+      "learning_rate": 7.63177683551939e-06,
+      "loss": 0.2754,
+      "step": 363
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6949155330657959,
+      "learning_rate": 7.618708799665745e-06,
+      "loss": 0.238,
+      "step": 364
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.761519730091095,
+      "learning_rate": 7.605616067315793e-06,
+      "loss": 0.2771,
+      "step": 365
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7326161861419678,
+      "learning_rate": 7.592498761944363e-06,
+      "loss": 0.177,
+      "step": 366
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5456448793411255,
+      "learning_rate": 7.579357007258022e-06,
+      "loss": 0.1321,
+      "step": 367
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6762471199035645,
+      "learning_rate": 7.56619092719392e-06,
+      "loss": 0.2127,
+      "step": 368
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.673669695854187,
+      "learning_rate": 7.5530006459186115e-06,
+      "loss": 0.1863,
+      "step": 369
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5625087022781372,
+      "learning_rate": 7.539786287826885e-06,
+      "loss": 0.1529,
+      "step": 370
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6047581434249878,
+      "learning_rate": 7.526547977540592e-06,
+      "loss": 0.1931,
+      "step": 371
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6393082737922668,
+      "learning_rate": 7.51328583990748e-06,
+      "loss": 0.2069,
+      "step": 372
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7750303745269775,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.2369,
+      "step": 373
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6093407273292542,
+      "learning_rate": 7.486690583114137e-06,
+      "loss": 0.185,
+      "step": 374
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7837387919425964,
+      "learning_rate": 7.473357714768222e-06,
+      "loss": 0.3191,
+      "step": 375
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8034994602203369,
+      "learning_rate": 7.460001520701756e-06,
+      "loss": 0.3518,
+      "step": 376
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7601478099822998,
+      "learning_rate": 7.446622126874219e-06,
+      "loss": 0.2954,
+      "step": 377
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7800700068473816,
+      "learning_rate": 7.4332196594638815e-06,
+      "loss": 0.2649,
+      "step": 378
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8061292171478271,
+      "learning_rate": 7.419794244866619e-06,
+      "loss": 0.2895,
+      "step": 379
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6951506733894348,
+      "learning_rate": 7.406346009694713e-06,
+      "loss": 0.2163,
+      "step": 380
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.575623631477356,
+      "learning_rate": 7.3928750807756656e-06,
+      "loss": 0.207,
+      "step": 381
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7265523672103882,
+      "learning_rate": 7.379381585150997e-06,
+      "loss": 0.2537,
+      "step": 382
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6455715298652649,
+      "learning_rate": 7.365865650075046e-06,
+      "loss": 0.2211,
+      "step": 383
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5819538831710815,
+      "learning_rate": 7.352327403013779e-06,
+      "loss": 0.1808,
+      "step": 384
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7239487171173096,
+      "learning_rate": 7.338766971643579e-06,
+      "loss": 0.2434,
+      "step": 385
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7268589735031128,
+      "learning_rate": 7.325184483850043e-06,
+      "loss": 0.2462,
+      "step": 386
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6222095489501953,
+      "learning_rate": 7.311580067726783e-06,
+      "loss": 0.192,
+      "step": 387
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6951655745506287,
+      "learning_rate": 7.297953851574207e-06,
+      "loss": 0.2471,
+      "step": 388
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8148622512817383,
+      "learning_rate": 7.284305963898315e-06,
+      "loss": 0.3615,
+      "step": 389
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6384522318840027,
+      "learning_rate": 7.270636533409491e-06,
+      "loss": 0.2832,
+      "step": 390
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7164002656936646,
+      "learning_rate": 7.25694568902128e-06,
+      "loss": 0.2781,
+      "step": 391
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.617064893245697,
+      "learning_rate": 7.243233559849179e-06,
+      "loss": 0.1868,
+      "step": 392
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.702542245388031,
+      "learning_rate": 7.229500275209418e-06,
+      "loss": 0.2169,
+      "step": 393
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7396453022956848,
+      "learning_rate": 7.215745964617737e-06,
+      "loss": 0.2465,
+      "step": 394
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7013355493545532,
+      "learning_rate": 7.201970757788172e-06,
+      "loss": 0.2857,
+      "step": 395
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6131812334060669,
+      "learning_rate": 7.188174784631824e-06,
+      "loss": 0.2012,
+      "step": 396
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7121132612228394,
+      "learning_rate": 7.174358175255636e-06,
+      "loss": 0.2295,
+      "step": 397
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6988712549209595,
+      "learning_rate": 7.160521059961169e-06,
+      "loss": 0.2304,
+      "step": 398
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5755978226661682,
+      "learning_rate": 7.14666356924337e-06,
+      "loss": 0.1861,
+      "step": 399
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7269181609153748,
+      "learning_rate": 7.132785833789344e-06,
+      "loss": 0.3132,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.24499449133872986,
+      "eval_runtime": 349.1649,
+      "eval_samples_per_second": 0.977,
+      "eval_steps_per_second": 0.246,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6201484799385071,
+      "learning_rate": 7.118887984477116e-06,
+      "loss": 0.2389,
+      "step": 401
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.629365086555481,
+      "learning_rate": 7.104970152374405e-06,
+      "loss": 0.2177,
+      "step": 402
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6488037109375,
+      "learning_rate": 7.091032468737382e-06,
+      "loss": 0.2393,
+      "step": 403
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8756319284439087,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": 0.2161,
+      "step": 404
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6746820211410522,
+      "learning_rate": 7.063098072819919e-06,
+      "loss": 0.2629,
+      "step": 405
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6890723705291748,
+      "learning_rate": 7.049101623982938e-06,
+      "loss": 0.2121,
+      "step": 406
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6225735545158386,
+      "learning_rate": 7.035085850496079e-06,
+      "loss": 0.2051,
+      "step": 407
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7883545756340027,
+      "learning_rate": 7.021050884539178e-06,
+      "loss": 0.2666,
+      "step": 408
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6250419020652771,
+      "learning_rate": 7.006996858473068e-06,
+      "loss": 0.2177,
+      "step": 409
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.678107738494873,
+      "learning_rate": 6.992923904838341e-06,
+      "loss": 0.2276,
+      "step": 410
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6031367778778076,
+      "learning_rate": 6.97883215635408e-06,
+      "loss": 0.2049,
+      "step": 411
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6895279288291931,
+      "learning_rate": 6.96472174591663e-06,
+      "loss": 0.2755,
+      "step": 412
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6867420077323914,
+      "learning_rate": 6.9505928065983275e-06,
+      "loss": 0.2522,
+      "step": 413
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5972843766212463,
+      "learning_rate": 6.936445471646249e-06,
+      "loss": 0.1986,
+      "step": 414
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6085368394851685,
+      "learning_rate": 6.922279874480959e-06,
+      "loss": 0.2766,
+      "step": 415
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6487507224082947,
+      "learning_rate": 6.908096148695251e-06,
+      "loss": 0.2687,
+      "step": 416
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6461302638053894,
+      "learning_rate": 6.893894428052881e-06,
+      "loss": 0.2268,
+      "step": 417
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6626706719398499,
+      "learning_rate": 6.879674846487314e-06,
+      "loss": 0.2386,
+      "step": 418
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7022949457168579,
+      "learning_rate": 6.865437538100456e-06,
+      "loss": 0.2843,
+      "step": 419
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6765186786651611,
+      "learning_rate": 6.8511826371613955e-06,
+      "loss": 0.1828,
+      "step": 420
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.661516547203064,
+      "learning_rate": 6.836910278105124e-06,
+      "loss": 0.2638,
+      "step": 421
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7024251818656921,
+      "learning_rate": 6.822620595531286e-06,
+      "loss": 0.2964,
+      "step": 422
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6266246438026428,
+      "learning_rate": 6.808313724202894e-06,
+      "loss": 0.2143,
+      "step": 423
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5504724979400635,
+      "learning_rate": 6.793989799045067e-06,
+      "loss": 0.1904,
+      "step": 424
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7041710019111633,
+      "learning_rate": 6.779648955143754e-06,
+      "loss": 0.2613,
+      "step": 425
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7407920360565186,
+      "learning_rate": 6.765291327744463e-06,
+      "loss": 0.2751,
+      "step": 426
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6743589043617249,
+      "learning_rate": 6.750917052250981e-06,
+      "loss": 0.2073,
+      "step": 427
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6536813378334045,
+      "learning_rate": 6.736526264224101e-06,
+      "loss": 0.2596,
+      "step": 428
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.669067919254303,
+      "learning_rate": 6.722119099380345e-06,
+      "loss": 0.2643,
+      "step": 429
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6680570840835571,
+      "learning_rate": 6.7076956935906756e-06,
+      "loss": 0.188,
+      "step": 430
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5632251501083374,
+      "learning_rate": 6.693256182879224e-06,
+      "loss": 0.1931,
+      "step": 431
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7128719091415405,
+      "learning_rate": 6.678800703422004e-06,
+      "loss": 0.3227,
+      "step": 432
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7528271675109863,
+      "learning_rate": 6.664329391545625e-06,
+      "loss": 0.3102,
+      "step": 433
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7777692079544067,
+      "learning_rate": 6.649842383726011e-06,
+      "loss": 0.3419,
+      "step": 434
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6767750978469849,
+      "learning_rate": 6.635339816587109e-06,
+      "loss": 0.2562,
+      "step": 435
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8037603497505188,
+      "learning_rate": 6.620821826899606e-06,
+      "loss": 0.3355,
+      "step": 436
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7408990859985352,
+      "learning_rate": 6.606288551579629e-06,
+      "loss": 0.2344,
+      "step": 437
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7058804035186768,
+      "learning_rate": 6.59174012768747e-06,
+      "loss": 0.3084,
+      "step": 438
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6666316390037537,
+      "learning_rate": 6.5771766924262795e-06,
+      "loss": 0.276,
+      "step": 439
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7084954977035522,
+      "learning_rate": 6.562598383140773e-06,
+      "loss": 0.2705,
+      "step": 440
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6333549618721008,
+      "learning_rate": 6.548005337315943e-06,
+      "loss": 0.2308,
+      "step": 441
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5120007991790771,
+      "learning_rate": 6.533397692575766e-06,
+      "loss": 0.1273,
+      "step": 442
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8390111327171326,
+      "learning_rate": 6.518775586681887e-06,
+      "loss": 0.3499,
+      "step": 443
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.647445797920227,
+      "learning_rate": 6.504139157532338e-06,
+      "loss": 0.2331,
+      "step": 444
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6378818154335022,
+      "learning_rate": 6.489488543160225e-06,
+      "loss": 0.1831,
+      "step": 445
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6484649777412415,
+      "learning_rate": 6.4748238817324395e-06,
+      "loss": 0.2106,
+      "step": 446
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.051329255104065,
+      "learning_rate": 6.460145311548341e-06,
+      "loss": 0.3166,
+      "step": 447
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7134484648704529,
+      "learning_rate": 6.445452971038464e-06,
+      "loss": 0.2727,
+      "step": 448
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7991112470626831,
+      "learning_rate": 6.430746998763204e-06,
+      "loss": 0.3293,
+      "step": 449
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7408881187438965,
+      "learning_rate": 6.41602753341152e-06,
+      "loss": 0.3266,
+      "step": 450
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9650523662567139,
+      "learning_rate": 6.4012947137996175e-06,
+      "loss": 0.3552,
+      "step": 451
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6928345561027527,
+      "learning_rate": 6.386548678869644e-06,
+      "loss": 0.2326,
+      "step": 452
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5790258646011353,
+      "learning_rate": 6.37178956768838e-06,
+      "loss": 0.1952,
+      "step": 453
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6764933466911316,
+      "learning_rate": 6.3570175194459205e-06,
+      "loss": 0.2527,
+      "step": 454
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6727444529533386,
+      "learning_rate": 6.342232673454371e-06,
+      "loss": 0.2522,
+      "step": 455
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5731246471405029,
+      "learning_rate": 6.3274351691465305e-06,
+      "loss": 0.1655,
+      "step": 456
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7887910604476929,
+      "learning_rate": 6.312625146074574e-06,
+      "loss": 0.2716,
+      "step": 457
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6423369646072388,
+      "learning_rate": 6.2978027439087405e-06,
+      "loss": 0.2245,
+      "step": 458
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7198876738548279,
+      "learning_rate": 6.28296810243601e-06,
+      "loss": 0.2578,
+      "step": 459
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5674285888671875,
+      "learning_rate": 6.268121361558792e-06,
+      "loss": 0.1801,
+      "step": 460
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7052492499351501,
+      "learning_rate": 6.2532626612936035e-06,
+      "loss": 0.3003,
+      "step": 461
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6291772127151489,
+      "learning_rate": 6.238392141769743e-06,
+      "loss": 0.1771,
+      "step": 462
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7111222743988037,
+      "learning_rate": 6.22350994322798e-06,
+      "loss": 0.3175,
+      "step": 463
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5738310217857361,
+      "learning_rate": 6.208616206019225e-06,
+      "loss": 0.1926,
+      "step": 464
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6483380198478699,
+      "learning_rate": 6.193711070603202e-06,
+      "loss": 0.2267,
+      "step": 465
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6434422135353088,
+      "learning_rate": 6.178794677547138e-06,
+      "loss": 0.215,
+      "step": 466
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7429325580596924,
+      "learning_rate": 6.163867167524419e-06,
+      "loss": 0.2847,
+      "step": 467
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.692653477191925,
+      "learning_rate": 6.14892868131328e-06,
+      "loss": 0.2467,
+      "step": 468
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.594670295715332,
+      "learning_rate": 6.1339793597954675e-06,
+      "loss": 0.187,
+      "step": 469
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7211235165596008,
+      "learning_rate": 6.119019343954914e-06,
+      "loss": 0.2295,
+      "step": 470
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6931540369987488,
+      "learning_rate": 6.104048774876407e-06,
+      "loss": 0.2345,
+      "step": 471
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7969695925712585,
+      "learning_rate": 6.089067793744258e-06,
+      "loss": 0.3256,
+      "step": 472
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8951885104179382,
+      "learning_rate": 6.074076541840978e-06,
+      "loss": 0.3994,
+      "step": 473
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5864188075065613,
+      "learning_rate": 6.059075160545933e-06,
+      "loss": 0.1807,
+      "step": 474
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7101827263832092,
+      "learning_rate": 6.044063791334023e-06,
+      "loss": 0.1999,
+      "step": 475
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.689575731754303,
+      "learning_rate": 6.029042575774334e-06,
+      "loss": 0.2511,
+      "step": 476
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7901589274406433,
+      "learning_rate": 6.01401165552882e-06,
+      "loss": 0.2774,
+      "step": 477
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7189447283744812,
+      "learning_rate": 5.998971172350953e-06,
+      "loss": 0.2859,
+      "step": 478
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6440539360046387,
+      "learning_rate": 5.9839212680843925e-06,
+      "loss": 0.1909,
+      "step": 479
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8534595370292664,
+      "learning_rate": 5.968862084661643e-06,
+      "loss": 0.3271,
+      "step": 480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5987100005149841,
+      "learning_rate": 5.9537937641027225e-06,
+      "loss": 0.2241,
+      "step": 481
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6581840515136719,
+      "learning_rate": 5.938716448513819e-06,
+      "loss": 0.2265,
+      "step": 482
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9717352390289307,
+      "learning_rate": 5.923630280085948e-06,
+      "loss": 0.371,
+      "step": 483
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7156874537467957,
+      "learning_rate": 5.908535401093618e-06,
+      "loss": 0.3291,
+      "step": 484
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6929928064346313,
+      "learning_rate": 5.893431953893483e-06,
+      "loss": 0.2386,
+      "step": 485
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6523422002792358,
+      "learning_rate": 5.878320080923001e-06,
+      "loss": 0.205,
+      "step": 486
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8255969882011414,
+      "learning_rate": 5.8631999246990954e-06,
+      "loss": 0.2516,
+      "step": 487
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8311260342597961,
+      "learning_rate": 5.848071627816804e-06,
+      "loss": 0.3562,
+      "step": 488
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7574479579925537,
+      "learning_rate": 5.832935332947937e-06,
+      "loss": 0.2694,
+      "step": 489
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6098451614379883,
+      "learning_rate": 5.817791182839734e-06,
+      "loss": 0.2251,
+      "step": 490
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7459378838539124,
+      "learning_rate": 5.8026393203135145e-06,
+      "loss": 0.2995,
+      "step": 491
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7321155071258545,
+      "learning_rate": 5.787479888263333e-06,
+      "loss": 0.2579,
+      "step": 492
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7155848145484924,
+      "learning_rate": 5.772313029654631e-06,
+      "loss": 0.2231,
+      "step": 493
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7132194638252258,
+      "learning_rate": 5.757138887522884e-06,
+      "loss": 0.2826,
+      "step": 494
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6572800874710083,
+      "learning_rate": 5.741957604972264e-06,
+      "loss": 0.2229,
+      "step": 495
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6408301591873169,
+      "learning_rate": 5.726769325174279e-06,
+      "loss": 0.237,
+      "step": 496
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7460103631019592,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 0.2454,
+      "step": 497
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7643551826477051,
+      "learning_rate": 5.696372346850842e-06,
+      "loss": 0.3264,
+      "step": 498
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6984466314315796,
+      "learning_rate": 5.68116393499295e-06,
+      "loss": 0.2581,
+      "step": 499
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5989054441452026,
+      "learning_rate": 5.66594909922011e-06,
+      "loss": 0.1823,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.24097025394439697,
+      "eval_runtime": 381.7737,
+      "eval_samples_per_second": 0.893,
+      "eval_steps_per_second": 0.225,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8489783406257629,
+      "learning_rate": 5.650727983020262e-06,
+      "loss": 0.3858,
+      "step": 501
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5319824814796448,
+      "learning_rate": 5.635500729940578e-06,
+      "loss": 0.1428,
+      "step": 502
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7632309198379517,
+      "learning_rate": 5.6202674835861045e-06,
+      "loss": 0.3202,
+      "step": 503
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7297234535217285,
+      "learning_rate": 5.605028387618412e-06,
+      "loss": 0.2656,
+      "step": 504
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.8261827230453491,
+      "learning_rate": 5.5897835857542315e-06,
+      "loss": 0.3094,
+      "step": 505
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6307250261306763,
+      "learning_rate": 5.574533221764109e-06,
+      "loss": 0.2165,
+      "step": 506
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6916859745979309,
+      "learning_rate": 5.559277439471047e-06,
+      "loss": 0.2101,
+      "step": 507
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6984620690345764,
+      "learning_rate": 5.544016382749146e-06,
+      "loss": 0.2769,
+      "step": 508
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6447324156761169,
+      "learning_rate": 5.528750195522244e-06,
+      "loss": 0.211,
+      "step": 509
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6922134757041931,
+      "learning_rate": 5.513479021762573e-06,
+      "loss": 0.2479,
+      "step": 510
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6607391238212585,
+      "learning_rate": 5.498203005489378e-06,
+      "loss": 0.1672,
+      "step": 511
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7346232533454895,
+      "learning_rate": 5.4829222907675895e-06,
+      "loss": 0.2102,
+      "step": 512
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6713115572929382,
+      "learning_rate": 5.467637021706438e-06,
+      "loss": 0.2303,
+      "step": 513
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8098154067993164,
+      "learning_rate": 5.4523473424581045e-06,
+      "loss": 0.2906,
+      "step": 514
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8032150864601135,
+      "learning_rate": 5.437053397216364e-06,
+      "loss": 0.2855,
+      "step": 515
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7085696458816528,
+      "learning_rate": 5.421755330215223e-06,
+      "loss": 0.2357,
+      "step": 516
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8494563102722168,
+      "learning_rate": 5.4064532857275645e-06,
+      "loss": 0.2773,
+      "step": 517
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7258232235908508,
+      "learning_rate": 5.3911474080637705e-06,
+      "loss": 0.2357,
+      "step": 518
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6795754432678223,
+      "learning_rate": 5.3758378415703825e-06,
+      "loss": 0.1905,
+      "step": 519
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8729014992713928,
+      "learning_rate": 5.3605247306287275e-06,
+      "loss": 0.3643,
+      "step": 520
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6851757764816284,
+      "learning_rate": 5.345208219653562e-06,
+      "loss": 0.2186,
+      "step": 521
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7230455875396729,
+      "learning_rate": 5.329888453091701e-06,
+      "loss": 0.2538,
+      "step": 522
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6683774590492249,
+      "learning_rate": 5.314565575420671e-06,
+      "loss": 0.1929,
+      "step": 523
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6849011778831482,
+      "learning_rate": 5.299239731147332e-06,
+      "loss": 0.1494,
+      "step": 524
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8139472603797913,
+      "learning_rate": 5.283911064806522e-06,
+      "loss": 0.2992,
+      "step": 525
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.709804117679596,
+      "learning_rate": 5.268579720959698e-06,
+      "loss": 0.2426,
+      "step": 526
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7699583172798157,
+      "learning_rate": 5.253245844193564e-06,
+      "loss": 0.2775,
+      "step": 527
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.880550742149353,
+      "learning_rate": 5.237909579118713e-06,
+      "loss": 0.274,
+      "step": 528
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7186005115509033,
+      "learning_rate": 5.222571070368258e-06,
+      "loss": 0.2439,
+      "step": 529
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6252779364585876,
+      "learning_rate": 5.2072304625964785e-06,
+      "loss": 0.1902,
+      "step": 530
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7325754761695862,
+      "learning_rate": 5.191887900477444e-06,
+      "loss": 0.2849,
+      "step": 531
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7283653616905212,
+      "learning_rate": 5.176543528703657e-06,
+      "loss": 0.241,
+      "step": 532
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6465723514556885,
+      "learning_rate": 5.161197491984684e-06,
+      "loss": 0.2358,
+      "step": 533
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5979306697845459,
+      "learning_rate": 5.1458499350458e-06,
+      "loss": 0.1647,
+      "step": 534
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8471933007240295,
+      "learning_rate": 5.130501002626609e-06,
+      "loss": 0.4086,
+      "step": 535
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6678841710090637,
+      "learning_rate": 5.11515083947969e-06,
+      "loss": 0.2123,
+      "step": 536
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7951354384422302,
+      "learning_rate": 5.099799590369231e-06,
+      "loss": 0.251,
+      "step": 537
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7659385800361633,
+      "learning_rate": 5.084447400069656e-06,
+      "loss": 0.305,
+      "step": 538
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.703631579875946,
+      "learning_rate": 5.069094413364272e-06,
+      "loss": 0.2384,
+      "step": 539
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8557327389717102,
+      "learning_rate": 5.053740775043891e-06,
+      "loss": 0.2994,
+      "step": 540
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.78694087266922,
+      "learning_rate": 5.038386629905475e-06,
+      "loss": 0.2705,
+      "step": 541
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7444816827774048,
+      "learning_rate": 5.0230321227507595e-06,
+      "loss": 0.2931,
+      "step": 542
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6637994050979614,
+      "learning_rate": 5.007677398384902e-06,
+      "loss": 0.2023,
+      "step": 543
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.795734167098999,
+      "learning_rate": 4.992322601615101e-06,
+      "loss": 0.2727,
+      "step": 544
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7429494261741638,
+      "learning_rate": 4.976967877249242e-06,
+      "loss": 0.2611,
+      "step": 545
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6305847764015198,
+      "learning_rate": 4.961613370094526e-06,
+      "loss": 0.2316,
+      "step": 546
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7047324776649475,
+      "learning_rate": 4.9462592249561095e-06,
+      "loss": 0.289,
+      "step": 547
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7027899622917175,
+      "learning_rate": 4.93090558663573e-06,
+      "loss": 0.1984,
+      "step": 548
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7407359480857849,
+      "learning_rate": 4.915552599930345e-06,
+      "loss": 0.2794,
+      "step": 549
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7558469772338867,
+      "learning_rate": 4.900200409630771e-06,
+      "loss": 0.2329,
+      "step": 550
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7046175599098206,
+      "learning_rate": 4.884849160520311e-06,
+      "loss": 0.2033,
+      "step": 551
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.762986421585083,
+      "learning_rate": 4.869498997373393e-06,
+      "loss": 0.2441,
+      "step": 552
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7733058333396912,
+      "learning_rate": 4.854150064954201e-06,
+      "loss": 0.2922,
+      "step": 553
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7784504294395447,
+      "learning_rate": 4.838802508015316e-06,
+      "loss": 0.2629,
+      "step": 554
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7874659895896912,
+      "learning_rate": 4.8234564712963445e-06,
+      "loss": 0.2723,
+      "step": 555
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6290090680122375,
+      "learning_rate": 4.808112099522558e-06,
+      "loss": 0.2316,
+      "step": 556
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.921485424041748,
+      "learning_rate": 4.792769537403523e-06,
+      "loss": 0.3212,
+      "step": 557
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8012232184410095,
+      "learning_rate": 4.777428929631743e-06,
+      "loss": 0.3222,
+      "step": 558
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8099982738494873,
+      "learning_rate": 4.762090420881289e-06,
+      "loss": 0.3166,
+      "step": 559
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.9401763677597046,
+      "learning_rate": 4.746754155806437e-06,
+      "loss": 0.3724,
+      "step": 560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6871545910835266,
+      "learning_rate": 4.731420279040303e-06,
+      "loss": 0.2512,
+      "step": 561
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6168978810310364,
+      "learning_rate": 4.716088935193479e-06,
+      "loss": 0.1568,
+      "step": 562
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7992426753044128,
+      "learning_rate": 4.700760268852669e-06,
+      "loss": 0.2947,
+      "step": 563
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5769720673561096,
+      "learning_rate": 4.68543442457933e-06,
+      "loss": 0.1636,
+      "step": 564
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6425333023071289,
+      "learning_rate": 4.670111546908299e-06,
+      "loss": 0.2066,
+      "step": 565
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7271445989608765,
+      "learning_rate": 4.65479178034644e-06,
+      "loss": 0.1569,
+      "step": 566
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7024662494659424,
+      "learning_rate": 4.639475269371273e-06,
+      "loss": 0.2402,
+      "step": 567
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8755499124526978,
+      "learning_rate": 4.624162158429618e-06,
+      "loss": 0.2204,
+      "step": 568
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8764632344245911,
+      "learning_rate": 4.608852591936231e-06,
+      "loss": 0.3024,
+      "step": 569
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7827917337417603,
+      "learning_rate": 4.593546714272438e-06,
+      "loss": 0.2817,
+      "step": 570
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8511632084846497,
+      "learning_rate": 4.5782446697847775e-06,
+      "loss": 0.3221,
+      "step": 571
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6528637409210205,
+      "learning_rate": 4.562946602783637e-06,
+      "loss": 0.1777,
+      "step": 572
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8783658146858215,
+      "learning_rate": 4.547652657541897e-06,
+      "loss": 0.4047,
+      "step": 573
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7391313910484314,
+      "learning_rate": 4.532362978293564e-06,
+      "loss": 0.2846,
+      "step": 574
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5948435068130493,
+      "learning_rate": 4.517077709232411e-06,
+      "loss": 0.1937,
+      "step": 575
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7957102656364441,
+      "learning_rate": 4.5017969945106225e-06,
+      "loss": 0.3197,
+      "step": 576
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.572628915309906,
+      "learning_rate": 4.486520978237431e-06,
+      "loss": 0.17,
+      "step": 577
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6679872274398804,
+      "learning_rate": 4.471249804477758e-06,
+      "loss": 0.1744,
+      "step": 578
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6934377551078796,
+      "learning_rate": 4.455983617250857e-06,
+      "loss": 0.2471,
+      "step": 579
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6016349792480469,
+      "learning_rate": 4.440722560528955e-06,
+      "loss": 0.1866,
+      "step": 580
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5812397003173828,
+      "learning_rate": 4.4254667782358925e-06,
+      "loss": 0.1788,
+      "step": 581
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7885786890983582,
+      "learning_rate": 4.410216414245771e-06,
+      "loss": 0.3005,
+      "step": 582
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7440980672836304,
+      "learning_rate": 4.394971612381591e-06,
+      "loss": 0.2512,
+      "step": 583
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6813129186630249,
+      "learning_rate": 4.379732516413897e-06,
+      "loss": 0.216,
+      "step": 584
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7254709005355835,
+      "learning_rate": 4.364499270059423e-06,
+      "loss": 0.249,
+      "step": 585
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7485551238059998,
+      "learning_rate": 4.34927201697974e-06,
+      "loss": 0.2576,
+      "step": 586
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6328112483024597,
+      "learning_rate": 4.334050900779893e-06,
+      "loss": 0.1902,
+      "step": 587
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8314217329025269,
+      "learning_rate": 4.318836065007052e-06,
+      "loss": 0.2998,
+      "step": 588
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6509906053543091,
+      "learning_rate": 4.303627653149159e-06,
+      "loss": 0.1919,
+      "step": 589
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7075145840644836,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": 0.2028,
+      "step": 590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7177186608314514,
+      "learning_rate": 4.2732306748257226e-06,
+      "loss": 0.2312,
+      "step": 591
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6522433161735535,
+      "learning_rate": 4.258042395027738e-06,
+      "loss": 0.1844,
+      "step": 592
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7007840871810913,
+      "learning_rate": 4.2428611124771184e-06,
+      "loss": 0.2224,
+      "step": 593
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7517430782318115,
+      "learning_rate": 4.227686970345373e-06,
+      "loss": 0.267,
+      "step": 594
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6153454184532166,
+      "learning_rate": 4.21252011173667e-06,
+      "loss": 0.2004,
+      "step": 595
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7708460092544556,
+      "learning_rate": 4.197360679686489e-06,
+      "loss": 0.2246,
+      "step": 596
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7786543369293213,
+      "learning_rate": 4.182208817160269e-06,
+      "loss": 0.3394,
+      "step": 597
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6301801800727844,
+      "learning_rate": 4.1670646670520656e-06,
+      "loss": 0.1996,
+      "step": 598
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7560685276985168,
+      "learning_rate": 4.151928372183198e-06,
+      "loss": 0.2335,
+      "step": 599
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.835745096206665,
+      "learning_rate": 4.136800075300906e-06,
+      "loss": 0.2274,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.23925696313381195,
+      "eval_runtime": 549.7058,
+      "eval_samples_per_second": 0.62,
+      "eval_steps_per_second": 0.156,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8754226565361023,
+      "learning_rate": 4.121679919077001e-06,
+      "loss": 0.3683,
+      "step": 601
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6908857226371765,
+      "learning_rate": 4.10656804610652e-06,
+      "loss": 0.203,
+      "step": 602
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.696713387966156,
+      "learning_rate": 4.091464598906385e-06,
+      "loss": 0.228,
+      "step": 603
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.676516592502594,
+      "learning_rate": 4.076369719914055e-06,
+      "loss": 0.2243,
+      "step": 604
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8648337721824646,
+      "learning_rate": 4.061283551486185e-06,
+      "loss": 0.3647,
+      "step": 605
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8069350719451904,
+      "learning_rate": 4.04620623589728e-06,
+      "loss": 0.3533,
+      "step": 606
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8564375638961792,
+      "learning_rate": 4.03113791533836e-06,
+      "loss": 0.3385,
+      "step": 607
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7426589131355286,
+      "learning_rate": 4.016078731915608e-06,
+      "loss": 0.2596,
+      "step": 608
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7388270497322083,
+      "learning_rate": 4.001028827649046e-06,
+      "loss": 0.2639,
+      "step": 609
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6917795538902283,
+      "learning_rate": 3.9859883444711795e-06,
+      "loss": 0.2463,
+      "step": 610
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.695612370967865,
+      "learning_rate": 3.970957424225666e-06,
+      "loss": 0.2193,
+      "step": 611
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6280950307846069,
+      "learning_rate": 3.955936208665979e-06,
+      "loss": 0.2037,
+      "step": 612
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.744050920009613,
+      "learning_rate": 3.940924839454067e-06,
+      "loss": 0.2406,
+      "step": 613
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8406124711036682,
+      "learning_rate": 3.925923458159023e-06,
+      "loss": 0.3304,
+      "step": 614
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7352477312088013,
+      "learning_rate": 3.910932206255742e-06,
+      "loss": 0.2496,
+      "step": 615
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6688184142112732,
+      "learning_rate": 3.895951225123595e-06,
+      "loss": 0.182,
+      "step": 616
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7015490531921387,
+      "learning_rate": 3.880980656045087e-06,
+      "loss": 0.2518,
+      "step": 617
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7293862700462341,
+      "learning_rate": 3.866020640204533e-06,
+      "loss": 0.2363,
+      "step": 618
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6916101574897766,
+      "learning_rate": 3.851071318686721e-06,
+      "loss": 0.2221,
+      "step": 619
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6372233629226685,
+      "learning_rate": 3.836132832475583e-06,
+      "loss": 0.1852,
+      "step": 620
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.899456262588501,
+      "learning_rate": 3.821205322452863e-06,
+      "loss": 0.2606,
+      "step": 621
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8286627531051636,
+      "learning_rate": 3.806288929396798e-06,
+      "loss": 0.3659,
+      "step": 622
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7856718301773071,
+      "learning_rate": 3.7913837939807763e-06,
+      "loss": 0.2654,
+      "step": 623
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.647342324256897,
+      "learning_rate": 3.77649005677202e-06,
+      "loss": 0.1858,
+      "step": 624
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8734769225120544,
+      "learning_rate": 3.7616078582302575e-06,
+      "loss": 0.35,
+      "step": 625
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7445953488349915,
+      "learning_rate": 3.7467373387063973e-06,
+      "loss": 0.2204,
+      "step": 626
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7629478573799133,
+      "learning_rate": 3.7318786384412076e-06,
+      "loss": 0.3257,
+      "step": 627
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.764764130115509,
+      "learning_rate": 3.7170318975639902e-06,
+      "loss": 0.2477,
+      "step": 628
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6604840159416199,
+      "learning_rate": 3.70219725609126e-06,
+      "loss": 0.2454,
+      "step": 629
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.714688241481781,
+      "learning_rate": 3.687374853925425e-06,
+      "loss": 0.2072,
+      "step": 630
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7985005974769592,
+      "learning_rate": 3.67256483085347e-06,
+      "loss": 0.2882,
+      "step": 631
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7973790764808655,
+      "learning_rate": 3.6577673265456296e-06,
+      "loss": 0.2818,
+      "step": 632
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.829279899597168,
+      "learning_rate": 3.6429824805540816e-06,
+      "loss": 0.3016,
+      "step": 633
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6215798258781433,
+      "learning_rate": 3.628210432311621e-06,
+      "loss": 0.1855,
+      "step": 634
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7925015091896057,
+      "learning_rate": 3.6134513211303555e-06,
+      "loss": 0.3206,
+      "step": 635
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7273108959197998,
+      "learning_rate": 3.5987052862003824e-06,
+      "loss": 0.2378,
+      "step": 636
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7947001457214355,
+      "learning_rate": 3.58397246658848e-06,
+      "loss": 0.3151,
+      "step": 637
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0714225769042969,
+      "learning_rate": 3.569253001236795e-06,
+      "loss": 0.2715,
+      "step": 638
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.784624457359314,
+      "learning_rate": 3.554547028961537e-06,
+      "loss": 0.2564,
+      "step": 639
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6738251447677612,
+      "learning_rate": 3.5398546884516606e-06,
+      "loss": 0.1875,
+      "step": 640
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7179457545280457,
+      "learning_rate": 3.5251761182675626e-06,
+      "loss": 0.2904,
+      "step": 641
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.740993082523346,
+      "learning_rate": 3.510511456839777e-06,
+      "loss": 0.2648,
+      "step": 642
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7999202013015747,
+      "learning_rate": 3.495860842467664e-06,
+      "loss": 0.2863,
+      "step": 643
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6961380839347839,
+      "learning_rate": 3.481224413318114e-06,
+      "loss": 0.2768,
+      "step": 644
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7229123115539551,
+      "learning_rate": 3.4666023074242356e-06,
+      "loss": 0.298,
+      "step": 645
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8005563020706177,
+      "learning_rate": 3.451994662684057e-06,
+      "loss": 0.3037,
+      "step": 646
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7264557480812073,
+      "learning_rate": 3.4374016168592296e-06,
+      "loss": 0.2452,
+      "step": 647
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8246781826019287,
+      "learning_rate": 3.4228233075737225e-06,
+      "loss": 0.2857,
+      "step": 648
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7802700996398926,
+      "learning_rate": 3.4082598723125303e-06,
+      "loss": 0.3002,
+      "step": 649
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6945613622665405,
+      "learning_rate": 3.393711448420372e-06,
+      "loss": 0.2183,
+      "step": 650
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7643357515335083,
+      "learning_rate": 3.379178173100396e-06,
+      "loss": 0.2716,
+      "step": 651
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6494430899620056,
+      "learning_rate": 3.3646601834128924e-06,
+      "loss": 0.174,
+      "step": 652
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7811049818992615,
+      "learning_rate": 3.3501576162739903e-06,
+      "loss": 0.2065,
+      "step": 653
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7266541123390198,
+      "learning_rate": 3.3356706084543766e-06,
+      "loss": 0.2717,
+      "step": 654
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5529820322990417,
+      "learning_rate": 3.3211992965779984e-06,
+      "loss": 0.1486,
+      "step": 655
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8447984457015991,
+      "learning_rate": 3.306743817120777e-06,
+      "loss": 0.2957,
+      "step": 656
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6964268684387207,
+      "learning_rate": 3.2923043064093252e-06,
+      "loss": 0.2753,
+      "step": 657
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7075220942497253,
+      "learning_rate": 3.2778809006196564e-06,
+      "loss": 0.2483,
+      "step": 658
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6105390787124634,
+      "learning_rate": 3.2634737357758994e-06,
+      "loss": 0.1621,
+      "step": 659
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6852737665176392,
+      "learning_rate": 3.2490829477490194e-06,
+      "loss": 0.2133,
+      "step": 660
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6731998920440674,
+      "learning_rate": 3.2347086722555382e-06,
+      "loss": 0.2016,
+      "step": 661
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6744810342788696,
+      "learning_rate": 3.220351044856247e-06,
+      "loss": 0.2303,
+      "step": 662
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6960051655769348,
+      "learning_rate": 3.206010200954935e-06,
+      "loss": 0.2265,
+      "step": 663
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7880598902702332,
+      "learning_rate": 3.191686275797107e-06,
+      "loss": 0.3208,
+      "step": 664
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7482811212539673,
+      "learning_rate": 3.177379404468715e-06,
+      "loss": 0.2257,
+      "step": 665
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5179324746131897,
+      "learning_rate": 3.1630897218948765e-06,
+      "loss": 0.1486,
+      "step": 666
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7205661535263062,
+      "learning_rate": 3.1488173628386066e-06,
+      "loss": 0.2822,
+      "step": 667
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7530344724655151,
+      "learning_rate": 3.1345624618995444e-06,
+      "loss": 0.2819,
+      "step": 668
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7841753959655762,
+      "learning_rate": 3.1203251535126867e-06,
+      "loss": 0.2162,
+      "step": 669
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7488176822662354,
+      "learning_rate": 3.10610557194712e-06,
+      "loss": 0.2414,
+      "step": 670
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8237447142601013,
+      "learning_rate": 3.0919038513047507e-06,
+      "loss": 0.3725,
+      "step": 671
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6755101084709167,
+      "learning_rate": 3.077720125519042e-06,
+      "loss": 0.2308,
+      "step": 672
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8368008732795715,
+      "learning_rate": 3.0635545283537523e-06,
+      "loss": 0.3018,
+      "step": 673
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6606637835502625,
+      "learning_rate": 3.0494071934016737e-06,
+      "loss": 0.2167,
+      "step": 674
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7647143006324768,
+      "learning_rate": 3.03527825408337e-06,
+      "loss": 0.3043,
+      "step": 675
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6645731329917908,
+      "learning_rate": 3.0211678436459214e-06,
+      "loss": 0.2039,
+      "step": 676
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7168276309967041,
+      "learning_rate": 3.007076095161662e-06,
+      "loss": 0.2454,
+      "step": 677
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7966533899307251,
+      "learning_rate": 2.9930031415269327e-06,
+      "loss": 0.2565,
+      "step": 678
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.594175398349762,
+      "learning_rate": 2.978949115460824e-06,
+      "loss": 0.152,
+      "step": 679
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7202569842338562,
+      "learning_rate": 2.9649141495039225e-06,
+      "loss": 0.2299,
+      "step": 680
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.818361222743988,
+      "learning_rate": 2.950898376017064e-06,
+      "loss": 0.2468,
+      "step": 681
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6550590395927429,
+      "learning_rate": 2.9369019271800827e-06,
+      "loss": 0.1583,
+      "step": 682
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6880379915237427,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 0.2737,
+      "step": 683
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6679242253303528,
+      "learning_rate": 2.908967531262618e-06,
+      "loss": 0.1807,
+      "step": 684
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7928236126899719,
+      "learning_rate": 2.895029847625595e-06,
+      "loss": 0.2728,
+      "step": 685
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9944189190864563,
+      "learning_rate": 2.8811120155228843e-06,
+      "loss": 0.3776,
+      "step": 686
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7912566661834717,
+      "learning_rate": 2.8672141662106577e-06,
+      "loss": 0.2994,
+      "step": 687
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8261896371841431,
+      "learning_rate": 2.8533364307566313e-06,
+      "loss": 0.271,
+      "step": 688
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8267972469329834,
+      "learning_rate": 2.839478940038833e-06,
+      "loss": 0.3058,
+      "step": 689
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7223398089408875,
+      "learning_rate": 2.8256418247443664e-06,
+      "loss": 0.256,
+      "step": 690
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6547335386276245,
+      "learning_rate": 2.811825215368179e-06,
+      "loss": 0.2153,
+      "step": 691
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8268930912017822,
+      "learning_rate": 2.7980292422118282e-06,
+      "loss": 0.3194,
+      "step": 692
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6824688911437988,
+      "learning_rate": 2.7842540353822634e-06,
+      "loss": 0.2299,
+      "step": 693
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7297709584236145,
+      "learning_rate": 2.770499724790584e-06,
+      "loss": 0.2297,
+      "step": 694
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7365825772285461,
+      "learning_rate": 2.7567664401508225e-06,
+      "loss": 0.2711,
+      "step": 695
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.870194673538208,
+      "learning_rate": 2.743054310978722e-06,
+      "loss": 0.3355,
+      "step": 696
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6595069766044617,
+      "learning_rate": 2.729363466590511e-06,
+      "loss": 0.1812,
+      "step": 697
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7013744711875916,
+      "learning_rate": 2.7156940361016864e-06,
+      "loss": 0.238,
+      "step": 698
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.745682954788208,
+      "learning_rate": 2.7020461484257952e-06,
+      "loss": 0.2412,
+      "step": 699
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.820225179195404,
+      "learning_rate": 2.6884199322732192e-06,
+      "loss": 0.3275,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.23962344229221344,
+      "eval_runtime": 340.0624,
+      "eval_samples_per_second": 1.003,
+      "eval_steps_per_second": 0.253,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8120321035385132,
+      "learning_rate": 2.6748155161499568e-06,
+      "loss": 0.2871,
+      "step": 701
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.643677294254303,
+      "learning_rate": 2.6612330283564226e-06,
+      "loss": 0.1843,
+      "step": 702
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7032471895217896,
+      "learning_rate": 2.6476725969862227e-06,
+      "loss": 0.1753,
+      "step": 703
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6656165719032288,
+      "learning_rate": 2.634134349924956e-06,
+      "loss": 0.1874,
+      "step": 704
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6445679068565369,
+      "learning_rate": 2.6206184148490066e-06,
+      "loss": 0.2454,
+      "step": 705
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7996844053268433,
+      "learning_rate": 2.6071249192243365e-06,
+      "loss": 0.279,
+      "step": 706
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7802110910415649,
+      "learning_rate": 2.5936539903052893e-06,
+      "loss": 0.318,
+      "step": 707
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8619838356971741,
+      "learning_rate": 2.580205755133384e-06,
+      "loss": 0.3339,
+      "step": 708
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7842369079589844,
+      "learning_rate": 2.5667803405361214e-06,
+      "loss": 0.3304,
+      "step": 709
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8101242184638977,
+      "learning_rate": 2.5533778731257824e-06,
+      "loss": 0.3354,
+      "step": 710
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7099646329879761,
+      "learning_rate": 2.5399984792982457e-06,
+      "loss": 0.2381,
+      "step": 711
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7360900640487671,
+      "learning_rate": 2.5266422852317796e-06,
+      "loss": 0.2642,
+      "step": 712
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7768506407737732,
+      "learning_rate": 2.513309416885865e-06,
+      "loss": 0.3074,
+      "step": 713
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8049039244651794,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.2184,
+      "step": 714
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6561914086341858,
+      "learning_rate": 2.4867141600925214e-06,
+      "loss": 0.2148,
+      "step": 715
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8116360902786255,
+      "learning_rate": 2.4734520224594094e-06,
+      "loss": 0.3025,
+      "step": 716
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.757860541343689,
+      "learning_rate": 2.4602137121731195e-06,
+      "loss": 0.2799,
+      "step": 717
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7722039222717285,
+      "learning_rate": 2.44699935408139e-06,
+      "loss": 0.2626,
+      "step": 718
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.52679443359375,
+      "learning_rate": 2.4338090728060808e-06,
+      "loss": 0.1255,
+      "step": 719
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6381962299346924,
+      "learning_rate": 2.4206429927419795e-06,
+      "loss": 0.2139,
+      "step": 720
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7625266313552856,
+      "learning_rate": 2.40750123805564e-06,
+      "loss": 0.2465,
+      "step": 721
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8177068829536438,
+      "learning_rate": 2.3943839326842096e-06,
+      "loss": 0.2712,
+      "step": 722
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6748345494270325,
+      "learning_rate": 2.381291200334257e-06,
+      "loss": 0.2079,
+      "step": 723
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8563744425773621,
+      "learning_rate": 2.368223164480611e-06,
+      "loss": 0.3,
+      "step": 724
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7199714779853821,
+      "learning_rate": 2.3551799483651894e-06,
+      "loss": 0.2661,
+      "step": 725
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.860863447189331,
+      "learning_rate": 2.342161674995843e-06,
+      "loss": 0.3407,
+      "step": 726
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7349135875701904,
+      "learning_rate": 2.3291684671451905e-06,
+      "loss": 0.2679,
+      "step": 727
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6539340615272522,
+      "learning_rate": 2.316200447349466e-06,
+      "loss": 0.2213,
+      "step": 728
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7105304002761841,
+      "learning_rate": 2.3032577379073577e-06,
+      "loss": 0.2413,
+      "step": 729
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.623697817325592,
+      "learning_rate": 2.2903404608788582e-06,
+      "loss": 0.1846,
+      "step": 730
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7190108299255371,
+      "learning_rate": 2.2774487380841116e-06,
+      "loss": 0.26,
+      "step": 731
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7020860910415649,
+      "learning_rate": 2.2645826911022656e-06,
+      "loss": 0.2464,
+      "step": 732
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7273522615432739,
+      "learning_rate": 2.2517424412703256e-06,
+      "loss": 0.2234,
+      "step": 733
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6023182272911072,
+      "learning_rate": 2.2389281096820077e-06,
+      "loss": 0.2128,
+      "step": 734
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6555721163749695,
+      "learning_rate": 2.2261398171865976e-06,
+      "loss": 0.2629,
+      "step": 735
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6597947478294373,
+      "learning_rate": 2.2133776843878185e-06,
+      "loss": 0.1795,
+      "step": 736
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7203525304794312,
+      "learning_rate": 2.2006418316426773e-06,
+      "loss": 0.2305,
+      "step": 737
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6759414076805115,
+      "learning_rate": 2.187932379060348e-06,
+      "loss": 0.2382,
+      "step": 738
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7028172612190247,
+      "learning_rate": 2.175249446501024e-06,
+      "loss": 0.1931,
+      "step": 739
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7945227026939392,
+      "learning_rate": 2.1625931535747964e-06,
+      "loss": 0.277,
+      "step": 740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7901543378829956,
+      "learning_rate": 2.1499636196405225e-06,
+      "loss": 0.2847,
+      "step": 741
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7968934178352356,
+      "learning_rate": 2.1373609638047033e-06,
+      "loss": 0.2271,
+      "step": 742
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7947246432304382,
+      "learning_rate": 2.1247853049203543e-06,
+      "loss": 0.2448,
+      "step": 743
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7870792150497437,
+      "learning_rate": 2.112236761585892e-06,
+      "loss": 0.2432,
+      "step": 744
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7615368962287903,
+      "learning_rate": 2.09971545214401e-06,
+      "loss": 0.265,
+      "step": 745
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8390629887580872,
+      "learning_rate": 2.087221494680563e-06,
+      "loss": 0.3023,
+      "step": 746
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7150559425354004,
+      "learning_rate": 2.074755007023461e-06,
+      "loss": 0.2097,
+      "step": 747
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5334554314613342,
+      "learning_rate": 2.0623161067415463e-06,
+      "loss": 0.1123,
+      "step": 748
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7979438304901123,
+      "learning_rate": 2.0499049111434922e-06,
+      "loss": 0.2728,
+      "step": 749
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7618849873542786,
+      "learning_rate": 2.0375215372766944e-06,
+      "loss": 0.2587,
+      "step": 750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.9019395112991333,
+      "learning_rate": 2.025166101926168e-06,
+      "loss": 0.3588,
+      "step": 751
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5421817302703857,
+      "learning_rate": 2.012838721613447e-06,
+      "loss": 0.1394,
+      "step": 752
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7142716646194458,
+      "learning_rate": 2.0005395125954814e-06,
+      "loss": 0.3148,
+      "step": 753
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6308157444000244,
+      "learning_rate": 1.988268590863546e-06,
+      "loss": 0.2071,
+      "step": 754
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7796697020530701,
+      "learning_rate": 1.9760260721421426e-06,
+      "loss": 0.2666,
+      "step": 755
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7846043109893799,
+      "learning_rate": 1.9638120718879133e-06,
+      "loss": 0.2956,
+      "step": 756
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.713147759437561,
+      "learning_rate": 1.951626705288544e-06,
+      "loss": 0.2859,
+      "step": 757
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.646619975566864,
+      "learning_rate": 1.9394700872616856e-06,
+      "loss": 0.1585,
+      "step": 758
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8083282113075256,
+      "learning_rate": 1.927342332453866e-06,
+      "loss": 0.3169,
+      "step": 759
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6538665890693665,
+      "learning_rate": 1.9152435552394105e-06,
+      "loss": 0.2333,
+      "step": 760
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6046538949012756,
+      "learning_rate": 1.9031738697193618e-06,
+      "loss": 0.1602,
+      "step": 761
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6781079769134521,
+      "learning_rate": 1.8911333897204071e-06,
+      "loss": 0.2269,
+      "step": 762
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7626420259475708,
+      "learning_rate": 1.8791222287937983e-06,
+      "loss": 0.2841,
+      "step": 763
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7151882648468018,
+      "learning_rate": 1.8671405002142918e-06,
+      "loss": 0.1745,
+      "step": 764
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7661588191986084,
+      "learning_rate": 1.855188316979068e-06,
+      "loss": 0.2579,
+      "step": 765
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7726594805717468,
+      "learning_rate": 1.8432657918066732e-06,
+      "loss": 0.2371,
+      "step": 766
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.79924476146698,
+      "learning_rate": 1.831373037135955e-06,
+      "loss": 0.3382,
+      "step": 767
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5692100524902344,
+      "learning_rate": 1.819510165125002e-06,
+      "loss": 0.1561,
+      "step": 768
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9645766615867615,
+      "learning_rate": 1.8076772876500831e-06,
+      "loss": 0.2402,
+      "step": 769
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6940447688102722,
+      "learning_rate": 1.7958745163045987e-06,
+      "loss": 0.261,
+      "step": 770
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7400657534599304,
+      "learning_rate": 1.7841019623980215e-06,
+      "loss": 0.2289,
+      "step": 771
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6116971969604492,
+      "learning_rate": 1.77235973695485e-06,
+      "loss": 0.168,
+      "step": 772
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7953826189041138,
+      "learning_rate": 1.760647950713566e-06,
+      "loss": 0.2794,
+      "step": 773
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6571024656295776,
+      "learning_rate": 1.7489667141255801e-06,
+      "loss": 0.2146,
+      "step": 774
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8735598921775818,
+      "learning_rate": 1.7373161373541968e-06,
+      "loss": 0.3331,
+      "step": 775
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.692879319190979,
+      "learning_rate": 1.7256963302735752e-06,
+      "loss": 0.2537,
+      "step": 776
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7424703240394592,
+      "learning_rate": 1.7141074024676913e-06,
+      "loss": 0.2342,
+      "step": 777
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6320103406906128,
+      "learning_rate": 1.702549463229305e-06,
+      "loss": 0.1437,
+      "step": 778
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7434752583503723,
+      "learning_rate": 1.6910226215589303e-06,
+      "loss": 0.2942,
+      "step": 779
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7751021385192871,
+      "learning_rate": 1.6795269861638041e-06,
+      "loss": 0.2776,
+      "step": 780
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.674355149269104,
+      "learning_rate": 1.6680626654568688e-06,
+      "loss": 0.1999,
+      "step": 781
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8401548862457275,
+      "learning_rate": 1.6566297675557392e-06,
+      "loss": 0.3487,
+      "step": 782
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8457766771316528,
+      "learning_rate": 1.6452284002816893e-06,
+      "loss": 0.2785,
+      "step": 783
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7536783814430237,
+      "learning_rate": 1.6338586711586358e-06,
+      "loss": 0.2188,
+      "step": 784
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7402835488319397,
+      "learning_rate": 1.6225206874121219e-06,
+      "loss": 0.2805,
+      "step": 785
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7845516800880432,
+      "learning_rate": 1.6112145559683057e-06,
+      "loss": 0.2494,
+      "step": 786
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8195182085037231,
+      "learning_rate": 1.5999403834529549e-06,
+      "loss": 0.3057,
+      "step": 787
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6405972242355347,
+      "learning_rate": 1.588698276190438e-06,
+      "loss": 0.2197,
+      "step": 788
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7488459348678589,
+      "learning_rate": 1.5774883402027208e-06,
+      "loss": 0.2605,
+      "step": 789
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8009891510009766,
+      "learning_rate": 1.5663106812083746e-06,
+      "loss": 0.2756,
+      "step": 790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.6462413668632507,
+      "learning_rate": 1.555165404621567e-06,
+      "loss": 0.2111,
+      "step": 791
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7553801536560059,
+      "learning_rate": 1.5440526155510766e-06,
+      "loss": 0.2879,
+      "step": 792
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8148736357688904,
+      "learning_rate": 1.5329724187992983e-06,
+      "loss": 0.3281,
+      "step": 793
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7564266324043274,
+      "learning_rate": 1.5219249188612556e-06,
+      "loss": 0.2453,
+      "step": 794
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7308322787284851,
+      "learning_rate": 1.5109102199236152e-06,
+      "loss": 0.2592,
+      "step": 795
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8379185199737549,
+      "learning_rate": 1.4999284258637054e-06,
+      "loss": 0.2998,
+      "step": 796
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7561792135238647,
+      "learning_rate": 1.488979640248534e-06,
+      "loss": 0.2468,
+      "step": 797
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6872095465660095,
+      "learning_rate": 1.4780639663338125e-06,
+      "loss": 0.217,
+      "step": 798
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5847635865211487,
+      "learning_rate": 1.467181507062987e-06,
+      "loss": 0.1827,
+      "step": 799
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6254853010177612,
+      "learning_rate": 1.4563323650662586e-06,
+      "loss": 0.1564,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.23893068730831146,
+      "eval_runtime": 334.7721,
+      "eval_samples_per_second": 1.019,
+      "eval_steps_per_second": 0.257,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6680090427398682,
+      "learning_rate": 1.4455166426596222e-06,
+      "loss": 0.1686,
+      "step": 801
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.9229381084442139,
+      "learning_rate": 1.434734441843899e-06,
+      "loss": 0.4003,
+      "step": 802
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6827018857002258,
+      "learning_rate": 1.4239858643037753e-06,
+      "loss": 0.2393,
+      "step": 803
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6769413352012634,
+      "learning_rate": 1.4132710114068427e-06,
+      "loss": 0.2169,
+      "step": 804
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6936076283454895,
+      "learning_rate": 1.4025899842026442e-06,
+      "loss": 0.1547,
+      "step": 805
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6686022281646729,
+      "learning_rate": 1.3919428834217163e-06,
+      "loss": 0.198,
+      "step": 806
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6601407527923584,
+      "learning_rate": 1.3813298094746491e-06,
+      "loss": 0.1894,
+      "step": 807
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6349535584449768,
+      "learning_rate": 1.3707508624511263e-06,
+      "loss": 0.18,
+      "step": 808
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7586642503738403,
+      "learning_rate": 1.3602061421189899e-06,
+      "loss": 0.2731,
+      "step": 809
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7111486196517944,
+      "learning_rate": 1.349695747923298e-06,
+      "loss": 0.2387,
+      "step": 810
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7135524749755859,
+      "learning_rate": 1.339219778985385e-06,
+      "loss": 0.2516,
+      "step": 811
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7469358444213867,
+      "learning_rate": 1.3287783341019278e-06,
+      "loss": 0.2263,
+      "step": 812
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8230369091033936,
+      "learning_rate": 1.3183715117440143e-06,
+      "loss": 0.2903,
+      "step": 813
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.83302241563797,
+      "learning_rate": 1.307999410056216e-06,
+      "loss": 0.2903,
+      "step": 814
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7573346495628357,
+      "learning_rate": 1.2976621268556571e-06,
+      "loss": 0.2714,
+      "step": 815
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.649545431137085,
+      "learning_rate": 1.2873597596311026e-06,
+      "loss": 0.2145,
+      "step": 816
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9165211915969849,
+      "learning_rate": 1.2770924055420258e-06,
+      "loss": 0.2811,
+      "step": 817
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.717332124710083,
+      "learning_rate": 1.2668601614177017e-06,
+      "loss": 0.3172,
+      "step": 818
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7683383822441101,
+      "learning_rate": 1.2566631237562894e-06,
+      "loss": 0.207,
+      "step": 819
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6939813494682312,
+      "learning_rate": 1.246501388723923e-06,
+      "loss": 0.2164,
+      "step": 820
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7078372836112976,
+      "learning_rate": 1.2363750521538064e-06,
+      "loss": 0.2329,
+      "step": 821
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8150206208229065,
+      "learning_rate": 1.2262842095453065e-06,
+      "loss": 0.3304,
+      "step": 822
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6956215500831604,
+      "learning_rate": 1.2162289560630524e-06,
+      "loss": 0.2567,
+      "step": 823
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8311361074447632,
+      "learning_rate": 1.2062093865360458e-06,
+      "loss": 0.3153,
+      "step": 824
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6261569261550903,
+      "learning_rate": 1.1962255954567537e-06,
+      "loss": 0.1892,
+      "step": 825
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7640008926391602,
+      "learning_rate": 1.1862776769802275e-06,
+      "loss": 0.2267,
+      "step": 826
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7890690565109253,
+      "learning_rate": 1.1763657249232107e-06,
+      "loss": 0.2429,
+      "step": 827
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6995456218719482,
+      "learning_rate": 1.1664898327632552e-06,
+      "loss": 0.1979,
+      "step": 828
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8067217469215393,
+      "learning_rate": 1.1566500936378389e-06,
+      "loss": 0.2707,
+      "step": 829
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7893016934394836,
+      "learning_rate": 1.146846600343488e-06,
+      "loss": 0.3009,
+      "step": 830
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.880754292011261,
+      "learning_rate": 1.1370794453349039e-06,
+      "loss": 0.3307,
+      "step": 831
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8914408087730408,
+      "learning_rate": 1.1273487207240845e-06,
+      "loss": 0.2634,
+      "step": 832
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5336819291114807,
+      "learning_rate": 1.1176545182794674e-06,
+      "loss": 0.1236,
+      "step": 833
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7172911167144775,
+      "learning_rate": 1.1079969294250515e-06,
+      "loss": 0.2377,
+      "step": 834
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8779481053352356,
+      "learning_rate": 1.0983760452395415e-06,
+      "loss": 0.2652,
+      "step": 835
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8408273458480835,
+      "learning_rate": 1.0887919564554893e-06,
+      "loss": 0.2963,
+      "step": 836
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8103682994842529,
+      "learning_rate": 1.079244753458437e-06,
+      "loss": 0.2976,
+      "step": 837
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7215389609336853,
+      "learning_rate": 1.0697345262860638e-06,
+      "loss": 0.3125,
+      "step": 838
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7416812181472778,
+      "learning_rate": 1.0602613646273374e-06,
+      "loss": 0.3058,
+      "step": 839
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8230550289154053,
+      "learning_rate": 1.0508253578216693e-06,
+      "loss": 0.3375,
+      "step": 840
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8321605324745178,
+      "learning_rate": 1.0414265948580694e-06,
+      "loss": 0.3159,
+      "step": 841
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7492278218269348,
+      "learning_rate": 1.0320651643743128e-06,
+      "loss": 0.2286,
+      "step": 842
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6947314143180847,
+      "learning_rate": 1.0227411546560962e-06,
+      "loss": 0.1957,
+      "step": 843
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7169224619865417,
+      "learning_rate": 1.0134546536362099e-06,
+      "loss": 0.2468,
+      "step": 844
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.794872522354126,
+      "learning_rate": 1.0042057488937067e-06,
+      "loss": 0.2482,
+      "step": 845
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8405777812004089,
+      "learning_rate": 9.949945276530782e-07,
+      "loss": 0.3541,
+      "step": 846
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6393839716911316,
+      "learning_rate": 9.858210767834292e-07,
+      "loss": 0.2073,
+      "step": 847
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8626486659049988,
+      "learning_rate": 9.76685482797662e-07,
+      "loss": 0.2682,
+      "step": 848
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.667521059513092,
+      "learning_rate": 9.675878318516546e-07,
+      "loss": 0.2127,
+      "step": 849
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9174708724021912,
+      "learning_rate": 9.58528209743459e-07,
+      "loss": 0.3151,
+      "step": 850
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6799070239067078,
+      "learning_rate": 9.495067019124793e-07,
+      "loss": 0.27,
+      "step": 851
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7855434417724609,
+      "learning_rate": 9.405233934386726e-07,
+      "loss": 0.2487,
+      "step": 852
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7704946398735046,
+      "learning_rate": 9.315783690417479e-07,
+      "loss": 0.2218,
+      "step": 853
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6957180500030518,
+      "learning_rate": 9.226717130803636e-07,
+      "loss": 0.2129,
+      "step": 854
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7627095580101013,
+      "learning_rate": 9.138035095513337e-07,
+      "loss": 0.2735,
+      "step": 855
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6550484895706177,
+      "learning_rate": 9.049738420888349e-07,
+      "loss": 0.229,
+      "step": 856
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7645315527915955,
+      "learning_rate": 8.961827939636198e-07,
+      "loss": 0.2348,
+      "step": 857
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5948243737220764,
+      "learning_rate": 8.874304480822271e-07,
+      "loss": 0.1143,
+      "step": 858
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6549612283706665,
+      "learning_rate": 8.787168869862067e-07,
+      "loss": 0.1665,
+      "step": 859
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7662985920906067,
+      "learning_rate": 8.700421928513353e-07,
+      "loss": 0.2988,
+      "step": 860
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.738470196723938,
+      "learning_rate": 8.614064474868423e-07,
+      "loss": 0.2364,
+      "step": 861
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7509949803352356,
+      "learning_rate": 8.528097323346408e-07,
+      "loss": 0.2596,
+      "step": 862
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7045740485191345,
+      "learning_rate": 8.442521284685573e-07,
+      "loss": 0.1838,
+      "step": 863
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6679222583770752,
+      "learning_rate": 8.357337165935675e-07,
+      "loss": 0.215,
+      "step": 864
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7682536840438843,
+      "learning_rate": 8.27254577045039e-07,
+      "loss": 0.2566,
+      "step": 865
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7528423070907593,
+      "learning_rate": 8.188147897879667e-07,
+      "loss": 0.2169,
+      "step": 866
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7058762907981873,
+      "learning_rate": 8.104144344162229e-07,
+      "loss": 0.1921,
+      "step": 867
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7042251825332642,
+      "learning_rate": 8.02053590151805e-07,
+      "loss": 0.239,
+      "step": 868
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7134466767311096,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 0.2534,
+      "step": 869
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.569417417049408,
+      "learning_rate": 7.854507499691006e-07,
+      "loss": 0.1404,
+      "step": 870
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.731052577495575,
+      "learning_rate": 7.772089106287345e-07,
+      "loss": 0.2561,
+      "step": 871
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8122794032096863,
+      "learning_rate": 7.690068955500623e-07,
+      "loss": 0.3047,
+      "step": 872
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6678743362426758,
+      "learning_rate": 7.608447820845771e-07,
+      "loss": 0.1985,
+      "step": 873
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6880447864532471,
+      "learning_rate": 7.527226472074678e-07,
+      "loss": 0.2263,
+      "step": 874
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7170695662498474,
+      "learning_rate": 7.446405675168938e-07,
+      "loss": 0.2498,
+      "step": 875
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7539134621620178,
+      "learning_rate": 7.365986192332624e-07,
+      "loss": 0.1666,
+      "step": 876
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7251168489456177,
+      "learning_rate": 7.285968781985093e-07,
+      "loss": 0.2157,
+      "step": 877
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5736756324768066,
+      "learning_rate": 7.206354198753862e-07,
+      "loss": 0.1379,
+      "step": 878
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.876884400844574,
+      "learning_rate": 7.127143193467445e-07,
+      "loss": 0.3385,
+      "step": 879
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7779631614685059,
+      "learning_rate": 7.048336513148307e-07,
+      "loss": 0.303,
+      "step": 880
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8376134037971497,
+      "learning_rate": 6.969934901005809e-07,
+      "loss": 0.3363,
+      "step": 881
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7117416262626648,
+      "learning_rate": 6.89193909642919e-07,
+      "loss": 0.2705,
+      "step": 882
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6771072149276733,
+      "learning_rate": 6.814349834980622e-07,
+      "loss": 0.2274,
+      "step": 883
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8444775342941284,
+      "learning_rate": 6.737167848388227e-07,
+      "loss": 0.3234,
+      "step": 884
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7159910798072815,
+      "learning_rate": 6.660393864539222e-07,
+      "loss": 0.2191,
+      "step": 885
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6470720171928406,
+      "learning_rate": 6.584028607473019e-07,
+      "loss": 0.1555,
+      "step": 886
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8595316410064697,
+      "learning_rate": 6.508072797374454e-07,
+      "loss": 0.305,
+      "step": 887
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7818759679794312,
+      "learning_rate": 6.432527150566903e-07,
+      "loss": 0.2768,
+      "step": 888
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7317049503326416,
+      "learning_rate": 6.3573923795056e-07,
+      "loss": 0.2567,
+      "step": 889
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7605761289596558,
+      "learning_rate": 6.282669192770896e-07,
+      "loss": 0.2877,
+      "step": 890
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7579327821731567,
+      "learning_rate": 6.208358295061572e-07,
+      "loss": 0.2585,
+      "step": 891
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6664072275161743,
+      "learning_rate": 6.134460387188207e-07,
+      "loss": 0.213,
+      "step": 892
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7787521481513977,
+      "learning_rate": 6.060976166066546e-07,
+      "loss": 0.2172,
+      "step": 893
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7663528919219971,
+      "learning_rate": 5.98790632471094e-07,
+      "loss": 0.2807,
+      "step": 894
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6902214884757996,
+      "learning_rate": 5.91525155222783e-07,
+      "loss": 0.2408,
+      "step": 895
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9152516722679138,
+      "learning_rate": 5.843012533809211e-07,
+      "loss": 0.3191,
+      "step": 896
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7439894676208496,
+      "learning_rate": 5.771189950726191e-07,
+      "loss": 0.2269,
+      "step": 897
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8530980348587036,
+      "learning_rate": 5.699784480322568e-07,
+      "loss": 0.3409,
+      "step": 898
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7167938351631165,
+      "learning_rate": 5.628796796008435e-07,
+      "loss": 0.2531,
+      "step": 899
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7698455452919006,
+      "learning_rate": 5.558227567253832e-07,
+      "loss": 0.2512,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.23878981173038483,
+      "eval_runtime": 335.0563,
+      "eval_samples_per_second": 1.018,
+      "eval_steps_per_second": 0.257,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7810540795326233,
+      "learning_rate": 5.488077459582425e-07,
+      "loss": 0.2887,
+      "step": 901
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7419719696044922,
+      "learning_rate": 5.418347134565249e-07,
+      "loss": 0.2431,
+      "step": 902
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7212926149368286,
+      "learning_rate": 5.349037249814443e-07,
+      "loss": 0.2713,
+      "step": 903
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.740754246711731,
+      "learning_rate": 5.28014845897708e-07,
+      "loss": 0.2437,
+      "step": 904
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7760072350502014,
+      "learning_rate": 5.211681411728969e-07,
+      "loss": 0.3061,
+      "step": 905
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7475609183311462,
+      "learning_rate": 5.14363675376855e-07,
+      "loss": 0.2418,
+      "step": 906
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6755508780479431,
+      "learning_rate": 5.076015126810784e-07,
+      "loss": 0.2594,
+      "step": 907
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7507731318473816,
+      "learning_rate": 5.008817168581137e-07,
+      "loss": 0.232,
+      "step": 908
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8247950077056885,
+      "learning_rate": 4.94204351280953e-07,
+      "loss": 0.275,
+      "step": 909
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7758000493049622,
+      "learning_rate": 4.875694789224372e-07,
+      "loss": 0.2721,
+      "step": 910
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7328912615776062,
+      "learning_rate": 4.809771623546627e-07,
+      "loss": 0.1906,
+      "step": 911
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6297558546066284,
+      "learning_rate": 4.7442746374839363e-07,
+      "loss": 0.1976,
+      "step": 912
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7924368977546692,
+      "learning_rate": 4.6792044487247003e-07,
+      "loss": 0.2643,
+      "step": 913
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6739974617958069,
+      "learning_rate": 4.614561670932288e-07,
+      "loss": 0.2081,
+      "step": 914
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.733456552028656,
+      "learning_rate": 4.5503469137392565e-07,
+      "loss": 0.2125,
+      "step": 915
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7064236998558044,
+      "learning_rate": 4.486560782741578e-07,
+      "loss": 0.1929,
+      "step": 916
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6808232069015503,
+      "learning_rate": 4.423203879492943e-07,
+      "loss": 0.1682,
+      "step": 917
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.9870191216468811,
+      "learning_rate": 4.360276801499086e-07,
+      "loss": 0.309,
+      "step": 918
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7222360968589783,
+      "learning_rate": 4.29778014221216e-07,
+      "loss": 0.2207,
+      "step": 919
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7426356673240662,
+      "learning_rate": 4.2357144910251003e-07,
+      "loss": 0.2496,
+      "step": 920
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7929947376251221,
+      "learning_rate": 4.1740804332661365e-07,
+      "loss": 0.2641,
+      "step": 921
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7465423345565796,
+      "learning_rate": 4.1128785501931947e-07,
+      "loss": 0.2528,
+      "step": 922
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6530787348747253,
+      "learning_rate": 4.05210941898847e-07,
+      "loss": 0.1902,
+      "step": 923
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5978170037269592,
+      "learning_rate": 3.9917736127529525e-07,
+      "loss": 0.1877,
+      "step": 924
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8100348114967346,
+      "learning_rate": 3.9318717005010496e-07,
+      "loss": 0.2846,
+      "step": 925
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7248678803443909,
+      "learning_rate": 3.8724042471551925e-07,
+      "loss": 0.2635,
+      "step": 926
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7893044352531433,
+      "learning_rate": 3.8133718135405283e-07,
+      "loss": 0.3017,
+      "step": 927
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6711196899414062,
+      "learning_rate": 3.7547749563796144e-07,
+      "loss": 0.2334,
+      "step": 928
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.797402024269104,
+      "learning_rate": 3.6966142282871873e-07,
+      "loss": 0.2679,
+      "step": 929
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7578075528144836,
+      "learning_rate": 3.638890177764948e-07,
+      "loss": 0.2445,
+      "step": 930
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7537961602210999,
+      "learning_rate": 3.581603349196372e-07,
+      "loss": 0.2642,
+      "step": 931
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7909368872642517,
+      "learning_rate": 3.524754282841575e-07,
+      "loss": 0.339,
+      "step": 932
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6665784120559692,
+      "learning_rate": 3.468343514832251e-07,
+      "loss": 0.2178,
+      "step": 933
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.672397255897522,
+      "learning_rate": 3.4123715771665786e-07,
+      "loss": 0.176,
+      "step": 934
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7691673040390015,
+      "learning_rate": 3.356838997704226e-07,
+      "loss": 0.3186,
+      "step": 935
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7448713183403015,
+      "learning_rate": 3.3017463001613625e-07,
+      "loss": 0.2278,
+      "step": 936
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6707292795181274,
+      "learning_rate": 3.247094004105711e-07,
+      "loss": 0.189,
+      "step": 937
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6711617708206177,
+      "learning_rate": 3.1928826249516984e-07,
+      "loss": 0.1772,
+      "step": 938
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7171783447265625,
+      "learning_rate": 3.1391126739555134e-07,
+      "loss": 0.2374,
+      "step": 939
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7142137885093689,
+      "learning_rate": 3.0857846582103504e-07,
+      "loss": 0.27,
+      "step": 940
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7926027774810791,
+      "learning_rate": 3.0328990806415935e-07,
+      "loss": 0.2567,
+      "step": 941
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8226751089096069,
+      "learning_rate": 2.9804564400021e-07,
+      "loss": 0.2975,
+      "step": 942
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.835600733757019,
+      "learning_rate": 2.92845723086746e-07,
+      "loss": 0.3123,
+      "step": 943
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7174431681632996,
+      "learning_rate": 2.876901943631372e-07,
+      "loss": 0.2407,
+      "step": 944
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.665255069732666,
+      "learning_rate": 2.8257910645009935e-07,
+      "loss": 0.2518,
+      "step": 945
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8344467878341675,
+      "learning_rate": 2.7751250754923574e-07,
+      "loss": 0.261,
+      "step": 946
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8135045766830444,
+      "learning_rate": 2.724904454425836e-07,
+      "loss": 0.3022,
+      "step": 947
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8465341329574585,
+      "learning_rate": 2.6751296749216395e-07,
+      "loss": 0.3494,
+      "step": 948
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.953181266784668,
+      "learning_rate": 2.625801206395312e-07,
+      "loss": 0.3343,
+      "step": 949
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7392805814743042,
+      "learning_rate": 2.5769195140533556e-07,
+      "loss": 0.2137,
+      "step": 950
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8524776101112366,
+      "learning_rate": 2.528485058888813e-07,
+      "loss": 0.3187,
+      "step": 951
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7331002354621887,
+      "learning_rate": 2.4804982976769197e-07,
+      "loss": 0.2828,
+      "step": 952
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9640606641769409,
+      "learning_rate": 2.4329596829708145e-07,
+      "loss": 0.3604,
+      "step": 953
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8314491510391235,
+      "learning_rate": 2.385869663097251e-07,
+      "loss": 0.2484,
+      "step": 954
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7713232040405273,
+      "learning_rate": 2.3392286821523723e-07,
+      "loss": 0.3094,
+      "step": 955
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7395336031913757,
+      "learning_rate": 2.2930371799975593e-07,
+      "loss": 0.2047,
+      "step": 956
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7347779273986816,
+      "learning_rate": 2.2472955922552164e-07,
+      "loss": 0.2186,
+      "step": 957
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8460179567337036,
+      "learning_rate": 2.202004350304715e-07,
+      "loss": 0.2971,
+      "step": 958
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8869367241859436,
+      "learning_rate": 2.1571638812783125e-07,
+      "loss": 0.3091,
+      "step": 959
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8569523692131042,
+      "learning_rate": 2.112774608057111e-07,
+      "loss": 0.259,
+      "step": 960
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7300274968147278,
+      "learning_rate": 2.068836949267089e-07,
+      "loss": 0.2597,
+      "step": 961
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7471948266029358,
+      "learning_rate": 2.0253513192751374e-07,
+      "loss": 0.2235,
+      "step": 962
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7210529446601868,
+      "learning_rate": 1.9823181281851513e-07,
+      "loss": 0.2374,
+      "step": 963
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7990643382072449,
+      "learning_rate": 1.9397377818341945e-07,
+      "loss": 0.2601,
+      "step": 964
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8323106169700623,
+      "learning_rate": 1.8976106817886197e-07,
+      "loss": 0.334,
+      "step": 965
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8063402771949768,
+      "learning_rate": 1.8559372253403152e-07,
+      "loss": 0.2553,
+      "step": 966
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7140501737594604,
+      "learning_rate": 1.814717805502958e-07,
+      "loss": 0.2324,
+      "step": 967
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6649307012557983,
+      "learning_rate": 1.7739528110083003e-07,
+      "loss": 0.2295,
+      "step": 968
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7434363961219788,
+      "learning_rate": 1.7336426263024896e-07,
+      "loss": 0.2461,
+      "step": 969
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5857141613960266,
+      "learning_rate": 1.6937876315424707e-07,
+      "loss": 0.1591,
+      "step": 970
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6685411930084229,
+      "learning_rate": 1.6543882025923884e-07,
+      "loss": 0.2452,
+      "step": 971
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.775608479976654,
+      "learning_rate": 1.6154447110200256e-07,
+      "loss": 0.3133,
+      "step": 972
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7381249070167542,
+      "learning_rate": 1.5769575240933422e-07,
+      "loss": 0.2296,
+      "step": 973
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.0521715879440308,
+      "learning_rate": 1.5389270047769578e-07,
+      "loss": 0.3918,
+      "step": 974
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.767364501953125,
+      "learning_rate": 1.5013535117287648e-07,
+      "loss": 0.2897,
+      "step": 975
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.598799467086792,
+      "learning_rate": 1.4642373992965365e-07,
+      "loss": 0.1702,
+      "step": 976
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6769062280654907,
+      "learning_rate": 1.427579017514591e-07,
+      "loss": 0.2071,
+      "step": 977
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6977760791778564,
+      "learning_rate": 1.3913787121004717e-07,
+      "loss": 0.2158,
+      "step": 978
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.9182886481285095,
+      "learning_rate": 1.3556368244517116e-07,
+      "loss": 0.3066,
+      "step": 979
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6467658877372742,
+      "learning_rate": 1.3203536916425842e-07,
+      "loss": 0.2063,
+      "step": 980
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.747478723526001,
+      "learning_rate": 1.2855296464209687e-07,
+      "loss": 0.2604,
+      "step": 981
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8544607758522034,
+      "learning_rate": 1.2511650172051636e-07,
+      "loss": 0.3624,
+      "step": 982
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7253915667533875,
+      "learning_rate": 1.217260128080816e-07,
+      "loss": 0.2207,
+      "step": 983
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7420864701271057,
+      "learning_rate": 1.183815298797858e-07,
+      "loss": 0.2552,
+      "step": 984
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7849740982055664,
+      "learning_rate": 1.1508308447674977e-07,
+      "loss": 0.2586,
+      "step": 985
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8122086524963379,
+      "learning_rate": 1.1183070770592442e-07,
+      "loss": 0.2471,
+      "step": 986
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6283002495765686,
+      "learning_rate": 1.0862443023979651e-07,
+      "loss": 0.2434,
+      "step": 987
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6364373564720154,
+      "learning_rate": 1.0546428231609896e-07,
+      "loss": 0.2017,
+      "step": 988
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8534179329872131,
+      "learning_rate": 1.0235029373752758e-07,
+      "loss": 0.363,
+      "step": 989
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6653255820274353,
+      "learning_rate": 9.928249387145983e-08,
+      "loss": 0.2208,
+      "step": 990
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7650781869888306,
+      "learning_rate": 9.626091164967599e-08,
+      "loss": 0.2101,
+      "step": 991
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6862877607345581,
+      "learning_rate": 9.32855755680867e-08,
+      "loss": 0.1838,
+      "step": 992
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8111743927001953,
+      "learning_rate": 9.035651368646647e-08,
+      "loss": 0.2743,
+      "step": 993
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.931341290473938,
+      "learning_rate": 8.747375362818667e-08,
+      "loss": 0.3782,
+      "step": 994
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7730108499526978,
+      "learning_rate": 8.463732257995571e-08,
+      "loss": 0.2632,
+      "step": 995
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6024267673492432,
+      "learning_rate": 8.184724729156379e-08,
+      "loss": 0.1676,
+      "step": 996
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7430551648139954,
+      "learning_rate": 7.910355407562742e-08,
+      "loss": 0.231,
+      "step": 997
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7887012958526611,
+      "learning_rate": 7.640626880734581e-08,
+      "loss": 0.188,
+      "step": 998
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6349278688430786,
+      "learning_rate": 7.375541692425325e-08,
+      "loss": 0.1936,
+      "step": 999
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7530270218849182,
+      "learning_rate": 7.115102342598101e-08,
+      "loss": 0.2884,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.23875625431537628,
+      "eval_runtime": 336.9275,
+      "eval_samples_per_second": 1.012,
+      "eval_steps_per_second": 0.255,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7549338340759277,
+      "learning_rate": 6.859311287402081e-08,
+      "loss": 0.1935,
+      "step": 1001
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6894304752349854,
+      "learning_rate": 6.608170939149283e-08,
+      "loss": 0.163,
+      "step": 1002
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7224317193031311,
+      "learning_rate": 6.361683666291973e-08,
+      "loss": 0.2729,
+      "step": 1003
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9411391615867615,
+      "learning_rate": 6.119851793400188e-08,
+      "loss": 0.3002,
+      "step": 1004
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7301786541938782,
+      "learning_rate": 5.882677601139919e-08,
+      "loss": 0.193,
+      "step": 1005
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7360510230064392,
+      "learning_rate": 5.6501633262513454e-08,
+      "loss": 0.2268,
+      "step": 1006
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8280237913131714,
+      "learning_rate": 5.4223111615281935e-08,
+      "loss": 0.3276,
+      "step": 1007
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.780082643032074,
+      "learning_rate": 5.1991232557966344e-08,
+      "loss": 0.2664,
+      "step": 1008
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5786715149879456,
+      "learning_rate": 4.9806017138953053e-08,
+      "loss": 0.1433,
+      "step": 1009
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6797437071800232,
+      "learning_rate": 4.766748596655268e-08,
+      "loss": 0.1986,
+      "step": 1010
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7584928870201111,
+      "learning_rate": 4.55756592088058e-08,
+      "loss": 0.2975,
+      "step": 1011
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8755053281784058,
+      "learning_rate": 4.3530556593294194e-08,
+      "loss": 0.3136,
+      "step": 1012
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8384301066398621,
+      "learning_rate": 4.1532197406954357e-08,
+      "loss": 0.2797,
+      "step": 1013
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8307690620422363,
+      "learning_rate": 3.958060049589485e-08,
+      "loss": 0.3254,
+      "step": 1014
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6803908348083496,
+      "learning_rate": 3.767578426521923e-08,
+      "loss": 0.2112,
+      "step": 1015
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7312712073326111,
+      "learning_rate": 3.581776667885062e-08,
+      "loss": 0.2179,
+      "step": 1016
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6420474052429199,
+      "learning_rate": 3.40065652593663e-08,
+      "loss": 0.2204,
+      "step": 1017
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7175133228302002,
+      "learning_rate": 3.2242197087828944e-08,
+      "loss": 0.2561,
+      "step": 1018
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7940815091133118,
+      "learning_rate": 3.052467880362675e-08,
+      "loss": 0.3655,
+      "step": 1019
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6603592038154602,
+      "learning_rate": 2.8854026604315798e-08,
+      "loss": 0.2561,
+      "step": 1020
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7610006928443909,
+      "learning_rate": 2.723025624546849e-08,
+      "loss": 0.3093,
+      "step": 1021
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7235128879547119,
+      "learning_rate": 2.5653383040524228e-08,
+      "loss": 0.285,
+      "step": 1022
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7884017825126648,
+      "learning_rate": 2.4123421860645646e-08,
+      "loss": 0.2275,
+      "step": 1023
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8262652158737183,
+      "learning_rate": 2.264038713457706e-08,
+      "loss": 0.3359,
+      "step": 1024
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6753684878349304,
+      "learning_rate": 2.1204292848509557e-08,
+      "loss": 0.2222,
+      "step": 1025
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.831423282623291,
+      "learning_rate": 1.98151525459489e-08,
+      "loss": 0.2273,
+      "step": 1026
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6862256526947021,
+      "learning_rate": 1.8472979327587292e-08,
+      "loss": 0.1598,
+      "step": 1027
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7676153182983398,
+      "learning_rate": 1.7177785851180127e-08,
+      "loss": 0.2834,
+      "step": 1028
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8516740798950195,
+      "learning_rate": 1.5929584331427218e-08,
+      "loss": 0.2838,
+      "step": 1029
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9321697354316711,
+      "learning_rate": 1.4728386539856754e-08,
+      "loss": 0.3405,
+      "step": 1030
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9343318939208984,
+      "learning_rate": 1.3574203804713748e-08,
+      "loss": 0.3623,
+      "step": 1031
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7499484419822693,
+      "learning_rate": 1.2467047010855659e-08,
+      "loss": 0.2538,
+      "step": 1032
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7069028615951538,
+      "learning_rate": 1.1406926599646373e-08,
+      "loss": 0.2288,
+      "step": 1033
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7008345723152161,
+      "learning_rate": 1.0393852568860718e-08,
+      "loss": 0.2133,
+      "step": 1034
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6172625422477722,
+      "learning_rate": 9.427834472588992e-09,
+      "loss": 0.1254,
+      "step": 1035
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7789463996887207,
+      "learning_rate": 8.508881421145366e-09,
+      "loss": 0.2299,
+      "step": 1036
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7931715846061707,
+      "learning_rate": 7.637002080985167e-09,
+      "loss": 0.2224,
+      "step": 1037
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8010485172271729,
+      "learning_rate": 6.81220467461996e-09,
+      "loss": 0.2829,
+      "step": 1038
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7054150700569153,
+      "learning_rate": 6.034496980542037e-09,
+      "loss": 0.202,
+      "step": 1039
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7824262380599976,
+      "learning_rate": 5.303886333151154e-09,
+      "loss": 0.2324,
+      "step": 1040
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7127040028572083,
+      "learning_rate": 4.620379622682358e-09,
+      "loss": 0.2468,
+      "step": 1041
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7531187534332275,
+      "learning_rate": 3.983983295146599e-09,
+      "loss": 0.2103,
+      "step": 1042
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6774247288703918,
+      "learning_rate": 3.394703352263551e-09,
+      "loss": 0.2367,
+      "step": 1043
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8376418352127075,
+      "learning_rate": 2.8525453514099966e-09,
+      "loss": 0.2763,
+      "step": 1044
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6632216572761536,
+      "learning_rate": 2.3575144055643094e-09,
+      "loss": 0.1538,
+      "step": 1045
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6601029634475708,
+      "learning_rate": 1.9096151832609378e-09,
+      "loss": 0.1739,
+      "step": 1046
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7332072854042053,
+      "learning_rate": 1.5088519085437736e-09,
+      "loss": 0.26,
+      "step": 1047
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7107292413711548,
+      "learning_rate": 1.1552283609272962e-09,
+      "loss": 0.2047,
+      "step": 1048
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8022828698158264,
+      "learning_rate": 8.487478753615997e-10,
+      "loss": 0.2998,
+      "step": 1049
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8682080507278442,
+      "learning_rate": 5.894133422001957e-10,
+      "loss": 0.3446,
+      "step": 1050
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6473788619041443,
+      "learning_rate": 3.772272071722594e-10,
+      "loss": 0.1819,
+      "step": 1051
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.738074004650116,
+      "learning_rate": 2.1219147136264383e-10,
+      "loss": 0.2266,
+      "step": 1052
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.805428683757782,
+      "learning_rate": 9.43076911874563e-11,
+      "loss": 0.3078,
+      "step": 1053
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8224419355392456,
+      "learning_rate": 2.3576978384065585e-11,
+      "loss": 0.2958,
+      "step": 1054
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6432749032974243,
+      "learning_rate": 0.0,
+      "loss": 0.187,
+      "step": 1055
+    },
+    {
+      "epoch": 1.0,
+      "step": 1055,
+      "total_flos": 1.1458310767955149e+17,
+      "train_loss": 0.26917854962049503,
+      "train_runtime": 48943.092,
+      "train_samples_per_second": 0.69,
+      "train_steps_per_second": 0.022
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1055,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.1458310767955149e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}