diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/README.md b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..d4576fe074287232d3836bf69c21d3f2593290d9
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/README.md
@@ -0,0 +1,9 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+
+- PEFT 0.4.0
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/adapter_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4cc2d022398faf8c87500c5422eb8402f4fa55e7
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/adapter_config.json
@@ -0,0 +1,26 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "/data/wiedmann/hub/models--lmms-lab--llava-onevision-qwen2-7b-ov/snapshots/0b07bf7565e244cf4f39982249eafe8cd799d6dd",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 256,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "gate_proj",
+    "k_proj",
+    "o_proj",
+    "q_proj",
+    "down_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}
\ No newline at end of file
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/adapter_model.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..bcf87c93e5f3027ddb4f342b7660e533dfd980ac
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61b79230bd7ccf207e3dff21d342351161c3abd7042e3c30a76e36f0a4bed426
+size 692127130
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..d9989c3ebebe651b269c3202d5362ec1fbe695a5
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/config.json
@@ -0,0 +1,68 @@
+{
+  "_name_or_path": "/data/wiedmann/hub/models--lmms-lab--llava-onevision-qwen2-7b-ov/snapshots/0b07bf7565e244cf4f39982249eafe8cd799d6dd",
+  "add_faster_video": false,
+  "add_time_instruction": false,
+  "architectures": [
+    "LlavaQwenForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "faster_token_stride": 10,
+  "force_sample": false,
+  "hidden_act": "silu",
+  "hidden_size": 3584,
+  "image_aspect_ratio": "square",
+  "image_crop_resolution": null,
+  "image_grid_pinpoints": null,
+  "image_split_resolution": null,
+  "image_token_index": 151646,
+  "initializer_range": 0.02,
+  "intermediate_size": 18944,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "mm_hidden_size": 1152,
+  "mm_newline_position": "no_token",
+  "mm_patch_merge_type": "spatial_unpad",
+  "mm_projector_lr": null,
+  "mm_projector_type": "mlp2x_gelu",
+  "mm_resampler_type": "streaming_agg",
+  "mm_spatial_pool_mode": "bilinear",
+  "mm_spatial_pool_stride": null,
+  "mm_streaming_frames_per_chunk": 4,
+  "mm_streaming_input_dim": 1152,
+  "mm_streaming_mlp_ratio": 4.0,
+  "mm_streaming_num_heads": 8,
+  "mm_streaming_num_layers": 4,
+  "mm_streaming_num_state_tokens": 4096,
+  "mm_streaming_patches_per_frame": 729,
+  "mm_streaming_state_dim": 1152,
+  "mm_streaming_vision_chunk_size": 4,
+  "mm_tunable_parts": "mm_mlp_adapter,mm_vision_resampler",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "google/siglip-so400m-patch14-384",
+  "mm_vision_tower_lr": null,
+  "model_type": "llava_qwen",
+  "num_attention_heads": 28,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 4,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": 131072,
+  "tie_word_embeddings": false,
+  "tokenizer_model_max_length": 8192,
+  "tokenizer_padding_side": "right",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.40.0.dev0",
+  "use_cache": true,
+  "use_mm_proj": true,
+  "use_pos_skipping": false,
+  "use_sliding_window": false,
+  "vision_tower_pretrained": null,
+  "vocab_size": 152064
+}
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/generation_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..19a297221acb87418d4388a3decef2282c6d7316
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/generation_config.json
@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.05,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.40.0.dev0"
+}
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/non_lora_trainables.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..5408b342b406445b5d198cb53cf1ddea85308a04
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8bf5787cf62ec454e7a0d96f86cc7d52635db6b3f57a88c6321851d0fb5cf00
+size 216160380
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/trainer_state.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..c8a3b503ae71286f3ae2263e1f3ed59857763a01
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_lora_agg_fc4_s4096/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": 33.463253021240234,
+      "learning_rate": 3.125e-06,
+      "loss": 1.1453,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 26.780750274658203,
+      "learning_rate": 6.25e-06,
+      "loss": 1.0555,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 29.528182983398438,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 1.1406,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.668144226074219,
+      "learning_rate": 1.25e-05,
+      "loss": 1.0648,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 10.581782341003418,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.6921,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.904531002044678,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.6445,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.807795763015747,
+      "learning_rate": 2.1875e-05,
+      "loss": 0.5963,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.281471014022827,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5237,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.3787190914154053,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 0.4223,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1695269346237183,
+      "learning_rate": 3.125e-05,
+      "loss": 0.3849,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6980259418487549,
+      "learning_rate": 3.4375e-05,
+      "loss": 0.3298,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6831204295158386,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.2897,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7375361323356628,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.3563,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7940900325775146,
+      "learning_rate": 4.375e-05,
+      "loss": 0.3812,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7872915267944336,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.5325,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7645293474197388,
+      "learning_rate": 5e-05,
+      "loss": 0.3874,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5848857760429382,
+      "learning_rate": 5.3125000000000004e-05,
+      "loss": 0.4368,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5760071873664856,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 0.2832,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5355564951896667,
+      "learning_rate": 5.9375e-05,
+      "loss": 0.3557,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.582057535648346,
+      "learning_rate": 6.25e-05,
+      "loss": 0.2499,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5517708659172058,
+      "learning_rate": 6.562500000000001e-05,
+      "loss": 0.3792,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.41149774193763733,
+      "learning_rate": 6.875e-05,
+      "loss": 0.1674,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5109180212020874,
+      "learning_rate": 7.1875e-05,
+      "loss": 0.3037,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7872570157051086,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.4778,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.48174795508384705,
+      "learning_rate": 7.8125e-05,
+      "loss": 0.2261,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6141104102134705,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 0.3808,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6023070812225342,
+      "learning_rate": 8.4375e-05,
+      "loss": 0.3627,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5329498648643494,
+      "learning_rate": 8.75e-05,
+      "loss": 0.3352,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5181126594543457,
+      "learning_rate": 9.062500000000001e-05,
+      "loss": 0.3279,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5490639805793762,
+      "learning_rate": 9.375e-05,
+      "loss": 0.3005,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5164560675621033,
+      "learning_rate": 9.687500000000001e-05,
+      "loss": 0.3032,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.639187216758728,
+      "learning_rate": 0.0001,
+      "loss": 0.3262,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5882562398910522,
+      "learning_rate": 9.999976423021617e-05,
+      "loss": 0.3716,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5443987250328064,
+      "learning_rate": 9.999905692308813e-05,
+      "loss": 0.3782,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6356123089790344,
+      "learning_rate": 9.999787808528638e-05,
+      "loss": 0.487,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4409308135509491,
+      "learning_rate": 9.999622772792828e-05,
+      "loss": 0.2907,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46439898014068604,
+      "learning_rate": 9.9994105866578e-05,
+      "loss": 0.3692,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5262895822525024,
+      "learning_rate": 9.999151252124639e-05,
+      "loss": 0.3646,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42949140071868896,
+      "learning_rate": 9.998844771639073e-05,
+      "loss": 0.2871,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3939957320690155,
+      "learning_rate": 9.998491148091457e-05,
+      "loss": 0.3124,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5722458958625793,
+      "learning_rate": 9.99809038481674e-05,
+      "loss": 0.4201,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.48468154668807983,
+      "learning_rate": 9.997642485594436e-05,
+      "loss": 0.3333,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.422470360994339,
+      "learning_rate": 9.99714745464859e-05,
+      "loss": 0.3529,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4298754930496216,
+      "learning_rate": 9.996605296647737e-05,
+      "loss": 0.2808,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5193277597427368,
+      "learning_rate": 9.996016016704854e-05,
+      "loss": 0.5033,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3961217701435089,
+      "learning_rate": 9.995379620377318e-05,
+      "loss": 0.2726,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.37250861525535583,
+      "learning_rate": 9.99469611366685e-05,
+      "loss": 0.2509,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.454837828874588,
+      "learning_rate": 9.993965503019457e-05,
+      "loss": 0.4041,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41022467613220215,
+      "learning_rate": 9.99318779532538e-05,
+      "loss": 0.3358,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.47888967394828796,
+      "learning_rate": 9.992362997919016e-05,
+      "loss": 0.3284,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.48224809765815735,
+      "learning_rate": 9.991491118578856e-05,
+      "loss": 0.3779,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4392887055873871,
+      "learning_rate": 9.990572165527413e-05,
+      "loss": 0.3401,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4877416789531708,
+      "learning_rate": 9.98960614743114e-05,
+      "loss": 0.4311,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5582373738288879,
+      "learning_rate": 9.988593073400354e-05,
+      "loss": 0.4371,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45542365312576294,
+      "learning_rate": 9.987532952989145e-05,
+      "loss": 0.3781,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41332772374153137,
+      "learning_rate": 9.986425796195287e-05,
+      "loss": 0.2731,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.491731196641922,
+      "learning_rate": 9.985271613460144e-05,
+      "loss": 0.4244,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38137125968933105,
+      "learning_rate": 9.984070415668574e-05,
+      "loss": 0.2419,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5110546350479126,
+      "learning_rate": 9.982822214148821e-05,
+      "loss": 0.4189,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4357592463493347,
+      "learning_rate": 9.981527020672413e-05,
+      "loss": 0.3087,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4841492176055908,
+      "learning_rate": 9.980184847454052e-05,
+      "loss": 0.3994,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.45283472537994385,
+      "learning_rate": 9.978795707151492e-05,
+      "loss": 0.3151,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41662806272506714,
+      "learning_rate": 9.977359612865423e-05,
+      "loss": 0.3086,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.454145610332489,
+      "learning_rate": 9.975876578139355e-05,
+      "loss": 0.3436,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.47985681891441345,
+      "learning_rate": 9.974346616959476e-05,
+      "loss": 0.4485,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40050870180130005,
+      "learning_rate": 9.972769743754531e-05,
+      "loss": 0.2936,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3651273548603058,
+      "learning_rate": 9.971145973395684e-05,
+      "loss": 0.2421,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39864468574523926,
+      "learning_rate": 9.969475321196373e-05,
+      "loss": 0.2122,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.42498964071273804,
+      "learning_rate": 9.96775780291217e-05,
+      "loss": 0.3032,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.43476343154907227,
+      "learning_rate": 9.965993434740633e-05,
+      "loss": 0.2892,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5220826268196106,
+      "learning_rate": 9.96418223332115e-05,
+      "loss": 0.3547,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4142451882362366,
+      "learning_rate": 9.962324215734782e-05,
+      "loss": 0.2061,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5300136804580688,
+      "learning_rate": 9.960419399504106e-05,
+      "loss": 0.397,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4280524253845215,
+      "learning_rate": 9.958467802593046e-05,
+      "loss": 0.272,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4605485200881958,
+      "learning_rate": 9.956469443406706e-05,
+      "loss": 0.3125,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.38905322551727295,
+      "learning_rate": 9.954424340791196e-05,
+      "loss": 0.2923,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4849022328853607,
+      "learning_rate": 9.952332514033447e-05,
+      "loss": 0.3444,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4296628534793854,
+      "learning_rate": 9.950193982861047e-05,
+      "loss": 0.3938,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4485234320163727,
+      "learning_rate": 9.948008767442034e-05,
+      "loss": 0.3401,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.39634308218955994,
+      "learning_rate": 9.94577688838472e-05,
+      "loss": 0.2878,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4237906336784363,
+      "learning_rate": 9.943498366737487e-05,
+      "loss": 0.3612,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.43370065093040466,
+      "learning_rate": 9.941173223988602e-05,
+      "loss": 0.3126,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.46338796615600586,
+      "learning_rate": 9.938801482065998e-05,
+      "loss": 0.3395,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4615630507469177,
+      "learning_rate": 9.93638316333708e-05,
+      "loss": 0.3873,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.43473371863365173,
+      "learning_rate": 9.933918290608509e-05,
+      "loss": 0.3233,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.43656232953071594,
+      "learning_rate": 9.93140688712598e-05,
+      "loss": 0.3439,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47754278779029846,
+      "learning_rate": 9.928848976574019e-05,
+      "loss": 0.4151,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.39599156379699707,
+      "learning_rate": 9.926244583075748e-05,
+      "loss": 0.323,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.46682918071746826,
+      "learning_rate": 9.923593731192655e-05,
+      "loss": 0.3748,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3897883892059326,
+      "learning_rate": 9.920896445924373e-05,
+      "loss": 0.2878,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3685135245323181,
+      "learning_rate": 9.918152752708438e-05,
+      "loss": 0.1981,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4372778832912445,
+      "learning_rate": 9.915362677420044e-05,
+      "loss": 0.3632,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.44404837489128113,
+      "learning_rate": 9.912526246371814e-05,
+      "loss": 0.4275,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46910810470581055,
+      "learning_rate": 9.909643486313533e-05,
+      "loss": 0.3601,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.48296812176704407,
+      "learning_rate": 9.906714424431913e-05,
+      "loss": 0.3837,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.47754955291748047,
+      "learning_rate": 9.903739088350325e-05,
+      "loss": 0.2722,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5000470280647278,
+      "learning_rate": 9.900717506128541e-05,
+      "loss": 0.3067,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4227772057056427,
+      "learning_rate": 9.897649706262473e-05,
+      "loss": 0.3085,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5166612863540649,
+      "learning_rate": 9.894535717683902e-05,
+      "loss": 0.3799,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4290980398654938,
+      "learning_rate": 9.891375569760204e-05,
+      "loss": 0.3404,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.3133184015750885,
+      "eval_runtime": 380.7356,
+      "eval_samples_per_second": 0.896,
+      "eval_steps_per_second": 0.226,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45301157236099243,
+      "learning_rate": 9.888169292294076e-05,
+      "loss": 0.2953,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44437727332115173,
+      "learning_rate": 9.884916915523251e-05,
+      "loss": 0.2939,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5186764597892761,
+      "learning_rate": 9.881618470120215e-05,
+      "loss": 0.361,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.46811190247535706,
+      "learning_rate": 9.87827398719192e-05,
+      "loss": 0.3804,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45014673471450806,
+      "learning_rate": 9.874883498279484e-05,
+      "loss": 0.2712,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.43706876039505005,
+      "learning_rate": 9.871447035357903e-05,
+      "loss": 0.3206,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41039609909057617,
+      "learning_rate": 9.867964630835743e-05,
+      "loss": 0.2765,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3698734641075134,
+      "learning_rate": 9.864436317554831e-05,
+      "loss": 0.2762,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4115326404571533,
+      "learning_rate": 9.860862128789953e-05,
+      "loss": 0.2836,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4087127447128296,
+      "learning_rate": 9.857242098248542e-05,
+      "loss": 0.227,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43598195910453796,
+      "learning_rate": 9.853576260070347e-05,
+      "loss": 0.3293,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.40193742513656616,
+      "learning_rate": 9.849864648827125e-05,
+      "loss": 0.2869,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.46908333897590637,
+      "learning_rate": 9.846107299522304e-05,
+      "loss": 0.3822,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4960734248161316,
+      "learning_rate": 9.842304247590667e-05,
+      "loss": 0.4207,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41240015625953674,
+      "learning_rate": 9.838455528897998e-05,
+      "loss": 0.2707,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.439078152179718,
+      "learning_rate": 9.834561179740762e-05,
+      "loss": 0.2976,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44322070479393005,
+      "learning_rate": 9.830621236845754e-05,
+      "loss": 0.3033,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42314597964286804,
+      "learning_rate": 9.826635737369751e-05,
+      "loss": 0.3193,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4348198473453522,
+      "learning_rate": 9.82260471889917e-05,
+      "loss": 0.2672,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41488930583000183,
+      "learning_rate": 9.818528219449705e-05,
+      "loss": 0.2617,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4016762971878052,
+      "learning_rate": 9.814406277465968e-05,
+      "loss": 0.255,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.465854287147522,
+      "learning_rate": 9.810238931821138e-05,
+      "loss": 0.3383,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.47409752011299133,
+      "learning_rate": 9.806026221816581e-05,
+      "loss": 0.4048,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4302755892276764,
+      "learning_rate": 9.801768187181486e-05,
+      "loss": 0.3688,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.40446576476097107,
+      "learning_rate": 9.797464868072488e-05,
+      "loss": 0.2707,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4418666660785675,
+      "learning_rate": 9.793116305073292e-05,
+      "loss": 0.3407,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4534258246421814,
+      "learning_rate": 9.788722539194291e-05,
+      "loss": 0.3779,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4234429895877838,
+      "learning_rate": 9.784283611872169e-05,
+      "loss": 0.2993,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.42587193846702576,
+      "learning_rate": 9.77979956496953e-05,
+      "loss": 0.3516,
+      "step": 129
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.48947110772132874,
+      "learning_rate": 9.77527044077448e-05,
+      "loss": 0.4567,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.47243231534957886,
+      "learning_rate": 9.770696282000244e-05,
+      "loss": 0.3589,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4766747057437897,
+      "learning_rate": 9.766077131784763e-05,
+      "loss": 0.4019,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4046523869037628,
+      "learning_rate": 9.761413033690276e-05,
+      "loss": 0.304,
+      "step": 133
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.44009989500045776,
+      "learning_rate": 9.756704031702918e-05,
+      "loss": 0.3779,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.397026002407074,
+      "learning_rate": 9.751950170232309e-05,
+      "loss": 0.2839,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.41876229643821716,
+      "learning_rate": 9.74715149411112e-05,
+      "loss": 0.3297,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4566335380077362,
+      "learning_rate": 9.742308048594665e-05,
+      "loss": 0.3794,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.28635403513908386,
+      "learning_rate": 9.73741987936047e-05,
+      "loss": 0.1685,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.36062803864479065,
+      "learning_rate": 9.732487032507837e-05,
+      "loss": 0.2339,
+      "step": 139
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4867377281188965,
+      "learning_rate": 9.727509554557417e-05,
+      "loss": 0.3986,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4133400022983551,
+      "learning_rate": 9.722487492450764e-05,
+      "loss": 0.279,
+      "step": 141
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4777900278568268,
+      "learning_rate": 9.717420893549902e-05,
+      "loss": 0.3673,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4630371034145355,
+      "learning_rate": 9.712309805636863e-05,
+      "loss": 0.3068,
+      "step": 143
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4813340902328491,
+      "learning_rate": 9.707154276913256e-05,
+      "loss": 0.3767,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.45727020502090454,
+      "learning_rate": 9.701954355999792e-05,
+      "loss": 0.3657,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.493734210729599,
+      "learning_rate": 9.696710091935841e-05,
+      "loss": 0.3377,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40601471066474915,
+      "learning_rate": 9.691421534178966e-05,
+      "loss": 0.2839,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.42050978541374207,
+      "learning_rate": 9.686088732604449e-05,
+      "loss": 0.3068,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.34204694628715515,
+      "learning_rate": 9.68071173750483e-05,
+      "loss": 0.2205,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4636201560497284,
+      "learning_rate": 9.675290599589429e-05,
+      "loss": 0.3712,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4551067650318146,
+      "learning_rate": 9.669825369983865e-05,
+      "loss": 0.2859,
+      "step": 151
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4192691743373871,
+      "learning_rate": 9.664316100229578e-05,
+      "loss": 0.3364,
+      "step": 152
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5012570023536682,
+      "learning_rate": 9.658762842283343e-05,
+      "loss": 0.3635,
+      "step": 153
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3482036888599396,
+      "learning_rate": 9.653165648516776e-05,
+      "loss": 0.2561,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.516733705997467,
+      "learning_rate": 9.647524571715843e-05,
+      "loss": 0.375,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3909461498260498,
+      "learning_rate": 9.641839665080363e-05,
+      "loss": 0.2994,
+      "step": 156
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4149559438228607,
+      "learning_rate": 9.636110982223505e-05,
+      "loss": 0.2856,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.41729792952537537,
+      "learning_rate": 9.630338577171282e-05,
+      "loss": 0.3193,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3693937063217163,
+      "learning_rate": 9.624522504362039e-05,
+      "loss": 0.2963,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.44135749340057373,
+      "learning_rate": 9.618662818645948e-05,
+      "loss": 0.3345,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.33225393295288086,
+      "learning_rate": 9.612759575284483e-05,
+      "loss": 0.2135,
+      "step": 161
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4729803502559662,
+      "learning_rate": 9.606812829949896e-05,
+      "loss": 0.3095,
+      "step": 162
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4297609329223633,
+      "learning_rate": 9.600822638724705e-05,
+      "loss": 0.2822,
+      "step": 163
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39072513580322266,
+      "learning_rate": 9.594789058101153e-05,
+      "loss": 0.259,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3938648998737335,
+      "learning_rate": 9.588712144980681e-05,
+      "loss": 0.3187,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4420059025287628,
+      "learning_rate": 9.582591956673387e-05,
+      "loss": 0.3217,
+      "step": 166
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4301930367946625,
+      "learning_rate": 9.576428550897489e-05,
+      "loss": 0.2597,
+      "step": 167
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4522709250450134,
+      "learning_rate": 9.570221985778785e-05,
+      "loss": 0.3467,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4223420023918152,
+      "learning_rate": 9.563972319850092e-05,
+      "loss": 0.3119,
+      "step": 169
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.29287779331207275,
+      "learning_rate": 9.557679612050708e-05,
+      "loss": 0.2044,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4239848256111145,
+      "learning_rate": 9.551343921725843e-05,
+      "loss": 0.3667,
+      "step": 171
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5214807391166687,
+      "learning_rate": 9.544965308626074e-05,
+      "loss": 0.4566,
+      "step": 172
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.365704208612442,
+      "learning_rate": 9.538543832906773e-05,
+      "loss": 0.2585,
+      "step": 173
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4296222925186157,
+      "learning_rate": 9.53207955512753e-05,
+      "loss": 0.3084,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4618189334869385,
+      "learning_rate": 9.525572536251607e-05,
+      "loss": 0.4142,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42373570799827576,
+      "learning_rate": 9.519022837645337e-05,
+      "loss": 0.3347,
+      "step": 176
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4333430230617523,
+      "learning_rate": 9.512430521077565e-05,
+      "loss": 0.2899,
+      "step": 177
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41771209239959717,
+      "learning_rate": 9.505795648719048e-05,
+      "loss": 0.3355,
+      "step": 178
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4613341987133026,
+      "learning_rate": 9.499118283141887e-05,
+      "loss": 0.3668,
+      "step": 179
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39535215497016907,
+      "learning_rate": 9.492398487318921e-05,
+      "loss": 0.3202,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.38543298840522766,
+      "learning_rate": 9.485636324623146e-05,
+      "loss": 0.258,
+      "step": 181
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4161026179790497,
+      "learning_rate": 9.478831858827104e-05,
+      "loss": 0.3575,
+      "step": 182
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4082934856414795,
+      "learning_rate": 9.471985154102292e-05,
+      "loss": 0.3281,
+      "step": 183
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7053436040878296,
+      "learning_rate": 9.465096275018556e-05,
+      "loss": 0.4193,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4834393560886383,
+      "learning_rate": 9.458165286543476e-05,
+      "loss": 0.3248,
+      "step": 185
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4567403793334961,
+      "learning_rate": 9.451192254041758e-05,
+      "loss": 0.3406,
+      "step": 186
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5055704712867737,
+      "learning_rate": 9.444177243274618e-05,
+      "loss": 0.4453,
+      "step": 187
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.449666291475296,
+      "learning_rate": 9.437120320399158e-05,
+      "loss": 0.3484,
+      "step": 188
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3901447355747223,
+      "learning_rate": 9.430021551967744e-05,
+      "loss": 0.2771,
+      "step": 189
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38137927651405334,
+      "learning_rate": 9.422881004927381e-05,
+      "loss": 0.2751,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4116629660129547,
+      "learning_rate": 9.415698746619079e-05,
+      "loss": 0.3352,
+      "step": 191
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4050225019454956,
+      "learning_rate": 9.408474844777217e-05,
+      "loss": 0.2582,
+      "step": 192
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.49112826585769653,
+      "learning_rate": 9.401209367528908e-05,
+      "loss": 0.4763,
+      "step": 193
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.43285635113716125,
+      "learning_rate": 9.393902383393347e-05,
+      "loss": 0.313,
+      "step": 194
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5483641624450684,
+      "learning_rate": 9.386553961281179e-05,
+      "loss": 0.4274,
+      "step": 195
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.44654542207717896,
+      "learning_rate": 9.379164170493844e-05,
+      "loss": 0.3819,
+      "step": 196
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.43054330348968506,
+      "learning_rate": 9.371733080722911e-05,
+      "loss": 0.3407,
+      "step": 197
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3351651728153229,
+      "learning_rate": 9.36426076204944e-05,
+      "loss": 0.2238,
+      "step": 198
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.42180266976356506,
+      "learning_rate": 9.35674728494331e-05,
+      "loss": 0.3569,
+      "step": 199
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38647109270095825,
+      "learning_rate": 9.349192720262555e-05,
+      "loss": 0.3005,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.3077980577945709,
+      "eval_runtime": 366.183,
+      "eval_samples_per_second": 0.931,
+      "eval_steps_per_second": 0.235,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3987346291542053,
+      "learning_rate": 9.341597139252698e-05,
+      "loss": 0.3195,
+      "step": 201
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.42176109552383423,
+      "learning_rate": 9.333960613546079e-05,
+      "loss": 0.3033,
+      "step": 202
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.0265607833862305,
+      "learning_rate": 9.326283215161178e-05,
+      "loss": 0.2669,
+      "step": 203
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.36215072870254517,
+      "learning_rate": 9.31856501650194e-05,
+      "loss": 0.1649,
+      "step": 204
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4004576802253723,
+      "learning_rate": 9.310806090357082e-05,
+      "loss": 0.3607,
+      "step": 205
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.335162878036499,
+      "learning_rate": 9.303006509899418e-05,
+      "loss": 0.1813,
+      "step": 206
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5005160570144653,
+      "learning_rate": 9.29516634868517e-05,
+      "loss": 0.3452,
+      "step": 207
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5310973525047302,
+      "learning_rate": 9.287285680653254e-05,
+      "loss": 0.3778,
+      "step": 208
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4045255184173584,
+      "learning_rate": 9.279364580124614e-05,
+      "loss": 0.3068,
+      "step": 209
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4020616114139557,
+      "learning_rate": 9.27140312180149e-05,
+      "loss": 0.2323,
+      "step": 210
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5438828468322754,
+      "learning_rate": 9.263401380766739e-05,
+      "loss": 0.4122,
+      "step": 211
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.46666768193244934,
+      "learning_rate": 9.255359432483105e-05,
+      "loss": 0.3492,
+      "step": 212
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3864457905292511,
+      "learning_rate": 9.247277352792533e-05,
+      "loss": 0.2443,
+      "step": 213
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.460303395986557,
+      "learning_rate": 9.239155217915422e-05,
+      "loss": 0.3574,
+      "step": 214
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3808644711971283,
+      "learning_rate": 9.230993104449939e-05,
+      "loss": 0.2338,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.388194739818573,
+      "learning_rate": 9.222791089371265e-05,
+      "loss": 0.3003,
+      "step": 216
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3851172626018524,
+      "learning_rate": 9.214549250030899e-05,
+      "loss": 0.2727,
+      "step": 217
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.43984320759773254,
+      "learning_rate": 9.206267664155907e-05,
+      "loss": 0.3941,
+      "step": 218
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.344109445810318,
+      "learning_rate": 9.197946409848194e-05,
+      "loss": 0.2302,
+      "step": 219
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.36127954721450806,
+      "learning_rate": 9.189585565583778e-05,
+      "loss": 0.294,
+      "step": 220
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3838326930999756,
+      "learning_rate": 9.181185210212034e-05,
+      "loss": 0.3235,
+      "step": 221
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3746877908706665,
+      "learning_rate": 9.172745422954961e-05,
+      "loss": 0.2592,
+      "step": 222
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3898646831512451,
+      "learning_rate": 9.164266283406433e-05,
+      "loss": 0.341,
+      "step": 223
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4040355086326599,
+      "learning_rate": 9.155747871531444e-05,
+      "loss": 0.3201,
+      "step": 224
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4587627053260803,
+      "learning_rate": 9.14719026766536e-05,
+      "loss": 0.3696,
+      "step": 225
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.41339704394340515,
+      "learning_rate": 9.13859355251316e-05,
+      "loss": 0.2971,
+      "step": 226
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3781212270259857,
+      "learning_rate": 9.129957807148666e-05,
+      "loss": 0.228,
+      "step": 227
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4829799234867096,
+      "learning_rate": 9.121283113013795e-05,
+      "loss": 0.4147,
+      "step": 228
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.37346261739730835,
+      "learning_rate": 9.112569551917774e-05,
+      "loss": 0.2586,
+      "step": 229
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.49685466289520264,
+      "learning_rate": 9.103817206036382e-05,
+      "loss": 0.3773,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4737912714481354,
+      "learning_rate": 9.095026157911166e-05,
+      "loss": 0.3402,
+      "step": 231
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4259354770183563,
+      "learning_rate": 9.086196490448668e-05,
+      "loss": 0.2581,
+      "step": 232
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.47733500599861145,
+      "learning_rate": 9.077328286919638e-05,
+      "loss": 0.3678,
+      "step": 233
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.46275725960731506,
+      "learning_rate": 9.068421630958254e-05,
+      "loss": 0.3789,
+      "step": 234
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3487975299358368,
+      "learning_rate": 9.059476606561328e-05,
+      "loss": 0.2243,
+      "step": 235
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.42151376605033875,
+      "learning_rate": 9.050493298087523e-05,
+      "loss": 0.3443,
+      "step": 236
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3983291685581207,
+      "learning_rate": 9.041471790256542e-05,
+      "loss": 0.3131,
+      "step": 237
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3939558267593384,
+      "learning_rate": 9.032412168148345e-05,
+      "loss": 0.3145,
+      "step": 238
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4804461598396301,
+      "learning_rate": 9.02331451720234e-05,
+      "loss": 0.3436,
+      "step": 239
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.37899306416511536,
+      "learning_rate": 9.014178923216572e-05,
+      "loss": 0.2432,
+      "step": 240
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.39545777440071106,
+      "learning_rate": 9.005005472346924e-05,
+      "loss": 0.2551,
+      "step": 241
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4486713111400604,
+      "learning_rate": 8.995794251106295e-05,
+      "loss": 0.3904,
+      "step": 242
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.40727540850639343,
+      "learning_rate": 8.986545346363792e-05,
+      "loss": 0.3112,
+      "step": 243
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4090732932090759,
+      "learning_rate": 8.977258845343905e-05,
+      "loss": 0.3381,
+      "step": 244
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3778274655342102,
+      "learning_rate": 8.967934835625689e-05,
+      "loss": 0.2922,
+      "step": 245
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.487745463848114,
+      "learning_rate": 8.958573405141932e-05,
+      "loss": 0.3359,
+      "step": 246
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4807039499282837,
+      "learning_rate": 8.949174642178332e-05,
+      "loss": 0.3894,
+      "step": 247
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.46274158358573914,
+      "learning_rate": 8.939738635372664e-05,
+      "loss": 0.3285,
+      "step": 248
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.48083364963531494,
+      "learning_rate": 8.930265473713938e-05,
+      "loss": 0.3729,
+      "step": 249
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.445803165435791,
+      "learning_rate": 8.920755246541563e-05,
+      "loss": 0.3487,
+      "step": 250
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4009513854980469,
+      "learning_rate": 8.911208043544513e-05,
+      "loss": 0.3125,
+      "step": 251
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3825117349624634,
+      "learning_rate": 8.90162395476046e-05,
+      "loss": 0.2932,
+      "step": 252
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5089893937110901,
+      "learning_rate": 8.89200307057495e-05,
+      "loss": 0.4571,
+      "step": 253
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.44665971398353577,
+      "learning_rate": 8.882345481720533e-05,
+      "loss": 0.3981,
+      "step": 254
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.40285229682922363,
+      "learning_rate": 8.872651279275916e-05,
+      "loss": 0.3255,
+      "step": 255
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.48048463463783264,
+      "learning_rate": 8.862920554665098e-05,
+      "loss": 0.3369,
+      "step": 256
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4042266309261322,
+      "learning_rate": 8.853153399656513e-05,
+      "loss": 0.3097,
+      "step": 257
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.45143160223960876,
+      "learning_rate": 8.843349906362163e-05,
+      "loss": 0.3747,
+      "step": 258
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.47038349509239197,
+      "learning_rate": 8.833510167236747e-05,
+      "loss": 0.3988,
+      "step": 259
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3986800014972687,
+      "learning_rate": 8.823634275076791e-05,
+      "loss": 0.3237,
+      "step": 260
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4058660864830017,
+      "learning_rate": 8.813722323019773e-05,
+      "loss": 0.3722,
+      "step": 261
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3654986023902893,
+      "learning_rate": 8.803774404543247e-05,
+      "loss": 0.256,
+      "step": 262
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3873800039291382,
+      "learning_rate": 8.793790613463955e-05,
+      "loss": 0.3277,
+      "step": 263
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.32056131958961487,
+      "learning_rate": 8.783771043936948e-05,
+      "loss": 0.2249,
+      "step": 264
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.41199854016304016,
+      "learning_rate": 8.773715790454695e-05,
+      "loss": 0.2951,
+      "step": 265
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.36119264364242554,
+      "learning_rate": 8.763624947846195e-05,
+      "loss": 0.2216,
+      "step": 266
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4735191762447357,
+      "learning_rate": 8.753498611276079e-05,
+      "loss": 0.3551,
+      "step": 267
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.46854570508003235,
+      "learning_rate": 8.743336876243712e-05,
+      "loss": 0.4061,
+      "step": 268
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5409342050552368,
+      "learning_rate": 8.7331398385823e-05,
+      "loss": 0.4504,
+      "step": 269
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4903641641139984,
+      "learning_rate": 8.722907594457974e-05,
+      "loss": 0.3853,
+      "step": 270
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4551154375076294,
+      "learning_rate": 8.712640240368899e-05,
+      "loss": 0.388,
+      "step": 271
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.48955848813056946,
+      "learning_rate": 8.702337873144343e-05,
+      "loss": 0.314,
+      "step": 272
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3990979492664337,
+      "learning_rate": 8.692000589943785e-05,
+      "loss": 0.2664,
+      "step": 273
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.48416367173194885,
+      "learning_rate": 8.681628488255986e-05,
+      "loss": 0.3942,
+      "step": 274
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3934306204319,
+      "learning_rate": 8.671221665898073e-05,
+      "loss": 0.2763,
+      "step": 275
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4343326687812805,
+      "learning_rate": 8.660780221014616e-05,
+      "loss": 0.321,
+      "step": 276
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4313034117221832,
+      "learning_rate": 8.650304252076703e-05,
+      "loss": 0.3044,
+      "step": 277
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3852527141571045,
+      "learning_rate": 8.639793857881011e-05,
+      "loss": 0.2629,
+      "step": 278
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5369113087654114,
+      "learning_rate": 8.629249137548874e-05,
+      "loss": 0.3859,
+      "step": 279
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40108469128608704,
+      "learning_rate": 8.618670190525352e-05,
+      "loss": 0.2608,
+      "step": 280
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4296994209289551,
+      "learning_rate": 8.608057116578283e-05,
+      "loss": 0.2928,
+      "step": 281
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4810967743396759,
+      "learning_rate": 8.597410015797358e-05,
+      "loss": 0.4075,
+      "step": 282
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.39262303709983826,
+      "learning_rate": 8.586728988593158e-05,
+      "loss": 0.2324,
+      "step": 283
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4846031963825226,
+      "learning_rate": 8.576014135696226e-05,
+      "loss": 0.3225,
+      "step": 284
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.43109941482543945,
+      "learning_rate": 8.565265558156101e-05,
+      "loss": 0.2848,
+      "step": 285
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5158528685569763,
+      "learning_rate": 8.554483357340379e-05,
+      "loss": 0.3483,
+      "step": 286
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3798547387123108,
+      "learning_rate": 8.543667634933742e-05,
+      "loss": 0.2486,
+      "step": 287
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.40132808685302734,
+      "learning_rate": 8.532818492937014e-05,
+      "loss": 0.2957,
+      "step": 288
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4158332347869873,
+      "learning_rate": 8.521936033666187e-05,
+      "loss": 0.3045,
+      "step": 289
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41023731231689453,
+      "learning_rate": 8.511020359751466e-05,
+      "loss": 0.3134,
+      "step": 290
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3341232240200043,
+      "learning_rate": 8.500071574136295e-05,
+      "loss": 0.2097,
+      "step": 291
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.40475884079933167,
+      "learning_rate": 8.489089780076386e-05,
+      "loss": 0.2887,
+      "step": 292
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.44635140895843506,
+      "learning_rate": 8.478075081138745e-05,
+      "loss": 0.2932,
+      "step": 293
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.524023175239563,
+      "learning_rate": 8.467027581200702e-05,
+      "loss": 0.4102,
+      "step": 294
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4615676999092102,
+      "learning_rate": 8.455947384448924e-05,
+      "loss": 0.3603,
+      "step": 295
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3583497107028961,
+      "learning_rate": 8.444834595378434e-05,
+      "loss": 0.2522,
+      "step": 296
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.41791126132011414,
+      "learning_rate": 8.433689318791626e-05,
+      "loss": 0.3444,
+      "step": 297
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.388347327709198,
+      "learning_rate": 8.422511659797279e-05,
+      "loss": 0.2983,
+      "step": 298
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4118559658527374,
+      "learning_rate": 8.411301723809563e-05,
+      "loss": 0.2932,
+      "step": 299
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.40110471844673157,
+      "learning_rate": 8.400059616547047e-05,
+      "loss": 0.3287,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.30673500895500183,
+      "eval_runtime": 375.3124,
+      "eval_samples_per_second": 0.909,
+      "eval_steps_per_second": 0.229,
+      "step": 300
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4560369551181793,
+      "learning_rate": 8.388785444031695e-05,
+      "loss": 0.3961,
+      "step": 301
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3951934278011322,
+      "learning_rate": 8.377479312587879e-05,
+      "loss": 0.2732,
+      "step": 302
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.40783199667930603,
+      "learning_rate": 8.366141328841366e-05,
+      "loss": 0.3369,
+      "step": 303
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.35192570090293884,
+      "learning_rate": 8.354771599718312e-05,
+      "loss": 0.2672,
+      "step": 304
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4301880896091461,
+      "learning_rate": 8.343370232444261e-05,
+      "loss": 0.3534,
+      "step": 305
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3767264783382416,
+      "learning_rate": 8.331937334543132e-05,
+      "loss": 0.2966,
+      "step": 306
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5049082040786743,
+      "learning_rate": 8.320473013836196e-05,
+      "loss": 0.437,
+      "step": 307
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4428861737251282,
+      "learning_rate": 8.308977378441072e-05,
+      "loss": 0.4045,
+      "step": 308
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.39177918434143066,
+      "learning_rate": 8.297450536770697e-05,
+      "loss": 0.3005,
+      "step": 309
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4027613401412964,
+      "learning_rate": 8.28589259753231e-05,
+      "loss": 0.2898,
+      "step": 310
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4139741361141205,
+      "learning_rate": 8.274303669726426e-05,
+      "loss": 0.3351,
+      "step": 311
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3958606421947479,
+      "learning_rate": 8.262683862645804e-05,
+      "loss": 0.3188,
+      "step": 312
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.45420899987220764,
+      "learning_rate": 8.251033285874421e-05,
+      "loss": 0.3544,
+      "step": 313
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43479791283607483,
+      "learning_rate": 8.239352049286435e-05,
+      "loss": 0.3241,
+      "step": 314
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4194246232509613,
+      "learning_rate": 8.22764026304515e-05,
+      "loss": 0.3307,
+      "step": 315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.43938562273979187,
+      "learning_rate": 8.21589803760198e-05,
+      "loss": 0.3521,
+      "step": 316
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3898228406906128,
+      "learning_rate": 8.204125483695403e-05,
+      "loss": 0.3206,
+      "step": 317
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3948107063770294,
+      "learning_rate": 8.192322712349917e-05,
+      "loss": 0.2679,
+      "step": 318
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.41959473490715027,
+      "learning_rate": 8.180489834875e-05,
+      "loss": 0.3209,
+      "step": 319
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4300706684589386,
+      "learning_rate": 8.168626962864045e-05,
+      "loss": 0.2702,
+      "step": 320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.47403374314308167,
+      "learning_rate": 8.156734208193327e-05,
+      "loss": 0.3136,
+      "step": 321
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4132087528705597,
+      "learning_rate": 8.144811683020932e-05,
+      "loss": 0.2905,
+      "step": 322
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.44167277216911316,
+      "learning_rate": 8.132859499785707e-05,
+      "loss": 0.3521,
+      "step": 323
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.41563671827316284,
+      "learning_rate": 8.120877771206201e-05,
+      "loss": 0.3158,
+      "step": 324
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.490485817193985,
+      "learning_rate": 8.108866610279593e-05,
+      "loss": 0.3958,
+      "step": 325
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4370015859603882,
+      "learning_rate": 8.096826130280639e-05,
+      "loss": 0.3083,
+      "step": 326
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.45056894421577454,
+      "learning_rate": 8.08475644476059e-05,
+      "loss": 0.3635,
+      "step": 327
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4257476329803467,
+      "learning_rate": 8.072657667546136e-05,
+      "loss": 0.3722,
+      "step": 328
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.41485312581062317,
+      "learning_rate": 8.060529912738315e-05,
+      "loss": 0.3685,
+      "step": 329
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4337831437587738,
+      "learning_rate": 8.048373294711455e-05,
+      "loss": 0.334,
+      "step": 330
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4475369155406952,
+      "learning_rate": 8.036187928112087e-05,
+      "loss": 0.3277,
+      "step": 331
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4627508223056793,
+      "learning_rate": 8.023973927857857e-05,
+      "loss": 0.4024,
+      "step": 332
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5114705562591553,
+      "learning_rate": 8.011731409136454e-05,
+      "loss": 0.3933,
+      "step": 333
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.42795073986053467,
+      "learning_rate": 7.99946048740452e-05,
+      "loss": 0.2481,
+      "step": 334
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4443337619304657,
+      "learning_rate": 7.987161278386554e-05,
+      "loss": 0.3656,
+      "step": 335
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4000591039657593,
+      "learning_rate": 7.974833898073832e-05,
+      "loss": 0.3081,
+      "step": 336
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.44498422741889954,
+      "learning_rate": 7.962478462723307e-05,
+      "loss": 0.3659,
+      "step": 337
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.43330293893814087,
+      "learning_rate": 7.950095088856508e-05,
+      "loss": 0.3587,
+      "step": 338
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.34501388669013977,
+      "learning_rate": 7.937683893258454e-05,
+      "loss": 0.2222,
+      "step": 339
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.46937355399131775,
+      "learning_rate": 7.925244992976538e-05,
+      "loss": 0.4076,
+      "step": 340
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.38550448417663574,
+      "learning_rate": 7.912778505319436e-05,
+      "loss": 0.3258,
+      "step": 341
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.327744722366333,
+      "learning_rate": 7.900284547855991e-05,
+      "loss": 0.1952,
+      "step": 342
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5067157745361328,
+      "learning_rate": 7.88776323841411e-05,
+      "loss": 0.3678,
+      "step": 343
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5155147314071655,
+      "learning_rate": 7.875214695079647e-05,
+      "loss": 0.3887,
+      "step": 344
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.32889455556869507,
+      "learning_rate": 7.862639036195298e-05,
+      "loss": 0.2573,
+      "step": 345
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4895000159740448,
+      "learning_rate": 7.850036380359478e-05,
+      "loss": 0.3491,
+      "step": 346
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4311446249485016,
+      "learning_rate": 7.837406846425204e-05,
+      "loss": 0.3781,
+      "step": 347
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.44158583879470825,
+      "learning_rate": 7.824750553498976e-05,
+      "loss": 0.3901,
+      "step": 348
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4824921190738678,
+      "learning_rate": 7.812067620939652e-05,
+      "loss": 0.3533,
+      "step": 349
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.38175836205482483,
+      "learning_rate": 7.799358168357323e-05,
+      "loss": 0.2895,
+      "step": 350
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5063992142677307,
+      "learning_rate": 7.786622315612183e-05,
+      "loss": 0.3905,
+      "step": 351
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.38219672441482544,
+      "learning_rate": 7.773860182813403e-05,
+      "loss": 0.2667,
+      "step": 352
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.46051451563835144,
+      "learning_rate": 7.761071890317994e-05,
+      "loss": 0.3404,
+      "step": 353
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3926549553871155,
+      "learning_rate": 7.748257558729677e-05,
+      "loss": 0.3185,
+      "step": 354
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4141940772533417,
+      "learning_rate": 7.735417308897736e-05,
+      "loss": 0.363,
+      "step": 355
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.36950042843818665,
+      "learning_rate": 7.72255126191589e-05,
+      "loss": 0.3064,
+      "step": 356
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4921727180480957,
+      "learning_rate": 7.709659539121144e-05,
+      "loss": 0.3228,
+      "step": 357
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.39004653692245483,
+      "learning_rate": 7.696742262092644e-05,
+      "loss": 0.3004,
+      "step": 358
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.34995216131210327,
+      "learning_rate": 7.683799552650534e-05,
+      "loss": 0.2341,
+      "step": 359
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.33777573704719543,
+      "learning_rate": 7.67083153285481e-05,
+      "loss": 0.2406,
+      "step": 360
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.38085034489631653,
+      "learning_rate": 7.657838325004159e-05,
+      "loss": 0.2613,
+      "step": 361
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.39987149834632874,
+      "learning_rate": 7.644820051634812e-05,
+      "loss": 0.3248,
+      "step": 362
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3920733332633972,
+      "learning_rate": 7.63177683551939e-05,
+      "loss": 0.281,
+      "step": 363
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.42502251267433167,
+      "learning_rate": 7.618708799665744e-05,
+      "loss": 0.2799,
+      "step": 364
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.42110690474510193,
+      "learning_rate": 7.605616067315793e-05,
+      "loss": 0.3568,
+      "step": 365
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3763708770275116,
+      "learning_rate": 7.592498761944362e-05,
+      "loss": 0.2282,
+      "step": 366
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3094440996646881,
+      "learning_rate": 7.579357007258022e-05,
+      "loss": 0.1697,
+      "step": 367
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4086095690727234,
+      "learning_rate": 7.56619092719392e-05,
+      "loss": 0.2668,
+      "step": 368
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.38525277376174927,
+      "learning_rate": 7.553000645918611e-05,
+      "loss": 0.1965,
+      "step": 369
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.33631476759910583,
+      "learning_rate": 7.539786287826885e-05,
+      "loss": 0.215,
+      "step": 370
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3649137318134308,
+      "learning_rate": 7.526547977540592e-05,
+      "loss": 0.2197,
+      "step": 371
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3965761363506317,
+      "learning_rate": 7.51328583990748e-05,
+      "loss": 0.2569,
+      "step": 372
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4292144477367401,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.2883,
+      "step": 373
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3937142789363861,
+      "learning_rate": 7.486690583114136e-05,
+      "loss": 0.2152,
+      "step": 374
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.47856462001800537,
+      "learning_rate": 7.473357714768222e-05,
+      "loss": 0.3565,
+      "step": 375
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5392282605171204,
+      "learning_rate": 7.460001520701755e-05,
+      "loss": 0.4562,
+      "step": 376
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4877152740955353,
+      "learning_rate": 7.446622126874218e-05,
+      "loss": 0.3506,
+      "step": 377
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5079466104507446,
+      "learning_rate": 7.433219659463881e-05,
+      "loss": 0.3718,
+      "step": 378
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4490622282028198,
+      "learning_rate": 7.41979424486662e-05,
+      "loss": 0.3605,
+      "step": 379
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.43833351135253906,
+      "learning_rate": 7.406346009694713e-05,
+      "loss": 0.2801,
+      "step": 380
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37128227949142456,
+      "learning_rate": 7.392875080775665e-05,
+      "loss": 0.2813,
+      "step": 381
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.44319096207618713,
+      "learning_rate": 7.379381585150996e-05,
+      "loss": 0.3141,
+      "step": 382
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.38892075419425964,
+      "learning_rate": 7.365865650075046e-05,
+      "loss": 0.2854,
+      "step": 383
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.34119877219200134,
+      "learning_rate": 7.35232740301378e-05,
+      "loss": 0.2291,
+      "step": 384
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37848344445228577,
+      "learning_rate": 7.338766971643579e-05,
+      "loss": 0.3159,
+      "step": 385
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4131897985935211,
+      "learning_rate": 7.325184483850042e-05,
+      "loss": 0.3124,
+      "step": 386
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.36052918434143066,
+      "learning_rate": 7.311580067726783e-05,
+      "loss": 0.2284,
+      "step": 387
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.366252064704895,
+      "learning_rate": 7.297953851574206e-05,
+      "loss": 0.2843,
+      "step": 388
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4672321379184723,
+      "learning_rate": 7.284305963898314e-05,
+      "loss": 0.4555,
+      "step": 389
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.42332160472869873,
+      "learning_rate": 7.27063653340949e-05,
+      "loss": 0.3465,
+      "step": 390
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40471723675727844,
+      "learning_rate": 7.25694568902128e-05,
+      "loss": 0.3351,
+      "step": 391
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.26821169257164,
+      "learning_rate": 7.243233559849179e-05,
+      "loss": 0.2063,
+      "step": 392
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3289814591407776,
+      "learning_rate": 7.229500275209418e-05,
+      "loss": 0.2559,
+      "step": 393
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.38697782158851624,
+      "learning_rate": 7.215745964617737e-05,
+      "loss": 0.2859,
+      "step": 394
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3759215474128723,
+      "learning_rate": 7.201970757788172e-05,
+      "loss": 0.3393,
+      "step": 395
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3581491708755493,
+      "learning_rate": 7.188174784631824e-05,
+      "loss": 0.2313,
+      "step": 396
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.417523056268692,
+      "learning_rate": 7.174358175255636e-05,
+      "loss": 0.302,
+      "step": 397
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.39265212416648865,
+      "learning_rate": 7.160521059961169e-05,
+      "loss": 0.2815,
+      "step": 398
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.29833367466926575,
+      "learning_rate": 7.146663569243371e-05,
+      "loss": 0.2267,
+      "step": 399
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4501245319843292,
+      "learning_rate": 7.132785833789344e-05,
+      "loss": 0.3964,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.30142679810523987,
+      "eval_runtime": 365.1403,
+      "eval_samples_per_second": 0.934,
+      "eval_steps_per_second": 0.236,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3599187135696411,
+      "learning_rate": 7.118887984477116e-05,
+      "loss": 0.2853,
+      "step": 401
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3475850522518158,
+      "learning_rate": 7.104970152374405e-05,
+      "loss": 0.3091,
+      "step": 402
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3699881434440613,
+      "learning_rate": 7.091032468737381e-05,
+      "loss": 0.2859,
+      "step": 403
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38557907938957214,
+      "learning_rate": 7.077075065009433e-05,
+      "loss": 0.2955,
+      "step": 404
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.35887739062309265,
+      "learning_rate": 7.063098072819919e-05,
+      "loss": 0.3039,
+      "step": 405
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.33867308497428894,
+      "learning_rate": 7.049101623982937e-05,
+      "loss": 0.2377,
+      "step": 406
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3663266599178314,
+      "learning_rate": 7.03508585049608e-05,
+      "loss": 0.2585,
+      "step": 407
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.41916424036026,
+      "learning_rate": 7.021050884539177e-05,
+      "loss": 0.3306,
+      "step": 408
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.36374786496162415,
+      "learning_rate": 7.006996858473069e-05,
+      "loss": 0.2759,
+      "step": 409
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3680070638656616,
+      "learning_rate": 6.99292390483834e-05,
+      "loss": 0.241,
+      "step": 410
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3433855175971985,
+      "learning_rate": 6.97883215635408e-05,
+      "loss": 0.2719,
+      "step": 411
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4610750377178192,
+      "learning_rate": 6.964721745916631e-05,
+      "loss": 0.3335,
+      "step": 412
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3723713755607605,
+      "learning_rate": 6.950592806598327e-05,
+      "loss": 0.2779,
+      "step": 413
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.36106976866722107,
+      "learning_rate": 6.936445471646249e-05,
+      "loss": 0.2676,
+      "step": 414
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3820250928401947,
+      "learning_rate": 6.922279874480959e-05,
+      "loss": 0.3419,
+      "step": 415
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.39156970381736755,
+      "learning_rate": 6.908096148695251e-05,
+      "loss": 0.3231,
+      "step": 416
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.37532204389572144,
+      "learning_rate": 6.89389442805288e-05,
+      "loss": 0.2794,
+      "step": 417
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.416196346282959,
+      "learning_rate": 6.879674846487314e-05,
+      "loss": 0.3069,
+      "step": 418
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4259437918663025,
+      "learning_rate": 6.865437538100457e-05,
+      "loss": 0.3403,
+      "step": 419
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3552284836769104,
+      "learning_rate": 6.851182637161396e-05,
+      "loss": 0.2356,
+      "step": 420
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.45776355266571045,
+      "learning_rate": 6.836910278105124e-05,
+      "loss": 0.3001,
+      "step": 421
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4252793788909912,
+      "learning_rate": 6.822620595531286e-05,
+      "loss": 0.3448,
+      "step": 422
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.379827082157135,
+      "learning_rate": 6.808313724202894e-05,
+      "loss": 0.2824,
+      "step": 423
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3481929898262024,
+      "learning_rate": 6.793989799045066e-05,
+      "loss": 0.234,
+      "step": 424
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.44025713205337524,
+      "learning_rate": 6.779648955143754e-05,
+      "loss": 0.3188,
+      "step": 425
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.41857773065567017,
+      "learning_rate": 6.765291327744463e-05,
+      "loss": 0.3306,
+      "step": 426
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3440551161766052,
+      "learning_rate": 6.750917052250982e-05,
+      "loss": 0.2878,
+      "step": 427
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.38679778575897217,
+      "learning_rate": 6.736526264224101e-05,
+      "loss": 0.307,
+      "step": 428
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.41354113817214966,
+      "learning_rate": 6.722119099380345e-05,
+      "loss": 0.3469,
+      "step": 429
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.35654181241989136,
+      "learning_rate": 6.707695693590676e-05,
+      "loss": 0.2143,
+      "step": 430
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.33059337735176086,
+      "learning_rate": 6.693256182879225e-05,
+      "loss": 0.28,
+      "step": 431
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.41064077615737915,
+      "learning_rate": 6.678800703422003e-05,
+      "loss": 0.3628,
+      "step": 432
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4637055993080139,
+      "learning_rate": 6.664329391545626e-05,
+      "loss": 0.3942,
+      "step": 433
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4418768882751465,
+      "learning_rate": 6.64984238372601e-05,
+      "loss": 0.4342,
+      "step": 434
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.37139832973480225,
+      "learning_rate": 6.635339816587109e-05,
+      "loss": 0.2988,
+      "step": 435
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4656209647655487,
+      "learning_rate": 6.620821826899606e-05,
+      "loss": 0.3783,
+      "step": 436
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4219549298286438,
+      "learning_rate": 6.606288551579629e-05,
+      "loss": 0.341,
+      "step": 437
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4449864625930786,
+      "learning_rate": 6.59174012768747e-05,
+      "loss": 0.3888,
+      "step": 438
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3836785554885864,
+      "learning_rate": 6.577176692426279e-05,
+      "loss": 0.3387,
+      "step": 439
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.397189736366272,
+      "learning_rate": 6.562598383140772e-05,
+      "loss": 0.3252,
+      "step": 440
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4128594696521759,
+      "learning_rate": 6.548005337315944e-05,
+      "loss": 0.2986,
+      "step": 441
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3210209012031555,
+      "learning_rate": 6.533397692575766e-05,
+      "loss": 0.1825,
+      "step": 442
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4985373020172119,
+      "learning_rate": 6.518775586681887e-05,
+      "loss": 0.4305,
+      "step": 443
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3670847415924072,
+      "learning_rate": 6.504139157532337e-05,
+      "loss": 0.2785,
+      "step": 444
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3467373549938202,
+      "learning_rate": 6.489488543160224e-05,
+      "loss": 0.215,
+      "step": 445
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.372842013835907,
+      "learning_rate": 6.47482388173244e-05,
+      "loss": 0.2443,
+      "step": 446
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.42306026816368103,
+      "learning_rate": 6.460145311548341e-05,
+      "loss": 0.3746,
+      "step": 447
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4473777115345001,
+      "learning_rate": 6.445452971038464e-05,
+      "loss": 0.3266,
+      "step": 448
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.44850483536720276,
+      "learning_rate": 6.430746998763204e-05,
+      "loss": 0.3745,
+      "step": 449
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.42026421427726746,
+      "learning_rate": 6.416027533411519e-05,
+      "loss": 0.3791,
+      "step": 450
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6435627937316895,
+      "learning_rate": 6.401294713799618e-05,
+      "loss": 0.4384,
+      "step": 451
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4087873697280884,
+      "learning_rate": 6.386548678869644e-05,
+      "loss": 0.3047,
+      "step": 452
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.35255467891693115,
+      "learning_rate": 6.37178956768838e-05,
+      "loss": 0.2746,
+      "step": 453
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4143827259540558,
+      "learning_rate": 6.35701751944592e-05,
+      "loss": 0.3124,
+      "step": 454
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4050014019012451,
+      "learning_rate": 6.342232673454371e-05,
+      "loss": 0.3387,
+      "step": 455
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32101985812187195,
+      "learning_rate": 6.32743516914653e-05,
+      "loss": 0.2068,
+      "step": 456
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.43388253450393677,
+      "learning_rate": 6.312625146074575e-05,
+      "loss": 0.3481,
+      "step": 457
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3686617314815521,
+      "learning_rate": 6.297802743908741e-05,
+      "loss": 0.2685,
+      "step": 458
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3812083303928375,
+      "learning_rate": 6.282968102436009e-05,
+      "loss": 0.3012,
+      "step": 459
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3404408097267151,
+      "learning_rate": 6.268121361558792e-05,
+      "loss": 0.2371,
+      "step": 460
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3948306739330292,
+      "learning_rate": 6.253262661293604e-05,
+      "loss": 0.3437,
+      "step": 461
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3803846538066864,
+      "learning_rate": 6.238392141769743e-05,
+      "loss": 0.2472,
+      "step": 462
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41616877913475037,
+      "learning_rate": 6.22350994322798e-05,
+      "loss": 0.3929,
+      "step": 463
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3899347484111786,
+      "learning_rate": 6.208616206019224e-05,
+      "loss": 0.2651,
+      "step": 464
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3516891598701477,
+      "learning_rate": 6.193711070603202e-05,
+      "loss": 0.252,
+      "step": 465
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3760213553905487,
+      "learning_rate": 6.178794677547137e-05,
+      "loss": 0.2775,
+      "step": 466
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4192168712615967,
+      "learning_rate": 6.163867167524419e-05,
+      "loss": 0.3203,
+      "step": 467
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.42506685853004456,
+      "learning_rate": 6.14892868131328e-05,
+      "loss": 0.3019,
+      "step": 468
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3687051236629486,
+      "learning_rate": 6.133979359795468e-05,
+      "loss": 0.269,
+      "step": 469
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.39751169085502625,
+      "learning_rate": 6.119019343954914e-05,
+      "loss": 0.2888,
+      "step": 470
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3806651532649994,
+      "learning_rate": 6.104048774876406e-05,
+      "loss": 0.3028,
+      "step": 471
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4066057503223419,
+      "learning_rate": 6.0890677937442574e-05,
+      "loss": 0.3665,
+      "step": 472
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5235038995742798,
+      "learning_rate": 6.074076541840977e-05,
+      "loss": 0.4791,
+      "step": 473
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3706027567386627,
+      "learning_rate": 6.059075160545933e-05,
+      "loss": 0.2344,
+      "step": 474
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3802247643470764,
+      "learning_rate": 6.0440637913340224e-05,
+      "loss": 0.2356,
+      "step": 475
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3983289301395416,
+      "learning_rate": 6.029042575774334e-05,
+      "loss": 0.3061,
+      "step": 476
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4319424033164978,
+      "learning_rate": 6.0140116555288204e-05,
+      "loss": 0.3347,
+      "step": 477
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.36478477716445923,
+      "learning_rate": 5.998971172350953e-05,
+      "loss": 0.3165,
+      "step": 478
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3390160799026489,
+      "learning_rate": 5.9839212680843925e-05,
+      "loss": 0.2553,
+      "step": 479
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4566447138786316,
+      "learning_rate": 5.968862084661643e-05,
+      "loss": 0.3702,
+      "step": 480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3880060315132141,
+      "learning_rate": 5.953793764102722e-05,
+      "loss": 0.3036,
+      "step": 481
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.37386608123779297,
+      "learning_rate": 5.938716448513818e-05,
+      "loss": 0.2834,
+      "step": 482
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5270214080810547,
+      "learning_rate": 5.923630280085948e-05,
+      "loss": 0.4461,
+      "step": 483
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.43317246437072754,
+      "learning_rate": 5.908535401093618e-05,
+      "loss": 0.3725,
+      "step": 484
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.36065560579299927,
+      "learning_rate": 5.8934319538934824e-05,
+      "loss": 0.2927,
+      "step": 485
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35703349113464355,
+      "learning_rate": 5.878320080923001e-05,
+      "loss": 0.2882,
+      "step": 486
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4906555712223053,
+      "learning_rate": 5.863199924699095e-05,
+      "loss": 0.3023,
+      "step": 487
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4809858500957489,
+      "learning_rate": 5.848071627816803e-05,
+      "loss": 0.4206,
+      "step": 488
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4332084357738495,
+      "learning_rate": 5.8329353329479366e-05,
+      "loss": 0.3507,
+      "step": 489
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3436206877231598,
+      "learning_rate": 5.8177911828397336e-05,
+      "loss": 0.248,
+      "step": 490
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4135282337665558,
+      "learning_rate": 5.802639320313514e-05,
+      "loss": 0.3737,
+      "step": 491
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.41700220108032227,
+      "learning_rate": 5.787479888263333e-05,
+      "loss": 0.3275,
+      "step": 492
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.41458266973495483,
+      "learning_rate": 5.77231302965463e-05,
+      "loss": 0.2995,
+      "step": 493
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.39454445242881775,
+      "learning_rate": 5.757138887522884e-05,
+      "loss": 0.2905,
+      "step": 494
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3396157920360565,
+      "learning_rate": 5.7419576049722646e-05,
+      "loss": 0.2446,
+      "step": 495
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3580017685890198,
+      "learning_rate": 5.7267693251742794e-05,
+      "loss": 0.2847,
+      "step": 496
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.42971375584602356,
+      "learning_rate": 5.7115741913664264e-05,
+      "loss": 0.299,
+      "step": 497
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4157896637916565,
+      "learning_rate": 5.696372346850841e-05,
+      "loss": 0.3868,
+      "step": 498
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4022786617279053,
+      "learning_rate": 5.68116393499295e-05,
+      "loss": 0.3238,
+      "step": 499
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3441603481769562,
+      "learning_rate": 5.6659490992201094e-05,
+      "loss": 0.2606,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.29364022612571716,
+      "eval_runtime": 357.746,
+      "eval_samples_per_second": 0.953,
+      "eval_steps_per_second": 0.24,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.46425434947013855,
+      "learning_rate": 5.650727983020262e-05,
+      "loss": 0.4477,
+      "step": 501
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3376691937446594,
+      "learning_rate": 5.635500729940577e-05,
+      "loss": 0.2148,
+      "step": 502
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.44071897864341736,
+      "learning_rate": 5.620267483586105e-05,
+      "loss": 0.3864,
+      "step": 503
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.42105183005332947,
+      "learning_rate": 5.6050283876184116e-05,
+      "loss": 0.3627,
+      "step": 504
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4721972644329071,
+      "learning_rate": 5.5897835857542317e-05,
+      "loss": 0.3645,
+      "step": 505
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3868240714073181,
+      "learning_rate": 5.574533221764109e-05,
+      "loss": 0.2704,
+      "step": 506
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.35737547278404236,
+      "learning_rate": 5.5592774394710465e-05,
+      "loss": 0.2831,
+      "step": 507
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.439684122800827,
+      "learning_rate": 5.5440163827491454e-05,
+      "loss": 0.2889,
+      "step": 508
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.38347262144088745,
+      "learning_rate": 5.5287501955222444e-05,
+      "loss": 0.2709,
+      "step": 509
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3540700674057007,
+      "learning_rate": 5.513479021762572e-05,
+      "loss": 0.2905,
+      "step": 510
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3754826784133911,
+      "learning_rate": 5.4982030054893785e-05,
+      "loss": 0.2256,
+      "step": 511
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.38406115770339966,
+      "learning_rate": 5.482922290767589e-05,
+      "loss": 0.2596,
+      "step": 512
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3912668526172638,
+      "learning_rate": 5.467637021706438e-05,
+      "loss": 0.2871,
+      "step": 513
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5177938938140869,
+      "learning_rate": 5.452347342458104e-05,
+      "loss": 0.3808,
+      "step": 514
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4036959409713745,
+      "learning_rate": 5.437053397216364e-05,
+      "loss": 0.3083,
+      "step": 515
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.40842923521995544,
+      "learning_rate": 5.4217553302152237e-05,
+      "loss": 0.3204,
+      "step": 516
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4749913513660431,
+      "learning_rate": 5.406453285727564e-05,
+      "loss": 0.3534,
+      "step": 517
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.40588313341140747,
+      "learning_rate": 5.39114740806377e-05,
+      "loss": 0.2788,
+      "step": 518
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.34570369124412537,
+      "learning_rate": 5.3758378415703825e-05,
+      "loss": 0.2439,
+      "step": 519
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4906198978424072,
+      "learning_rate": 5.360524730628728e-05,
+      "loss": 0.4068,
+      "step": 520
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4042207598686218,
+      "learning_rate": 5.345208219653561e-05,
+      "loss": 0.2866,
+      "step": 521
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4069378077983856,
+      "learning_rate": 5.3298884530917014e-05,
+      "loss": 0.3459,
+      "step": 522
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3647920489311218,
+      "learning_rate": 5.3145655754206714e-05,
+      "loss": 0.2581,
+      "step": 523
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3191315233707428,
+      "learning_rate": 5.2992397311473316e-05,
+      "loss": 0.2091,
+      "step": 524
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4241507947444916,
+      "learning_rate": 5.2839110648065214e-05,
+      "loss": 0.3252,
+      "step": 525
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.44086626172065735,
+      "learning_rate": 5.2685797209596976e-05,
+      "loss": 0.3245,
+      "step": 526
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4368096888065338,
+      "learning_rate": 5.2532458441935636e-05,
+      "loss": 0.3158,
+      "step": 527
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.44598886370658875,
+      "learning_rate": 5.2379095791187124e-05,
+      "loss": 0.3475,
+      "step": 528
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3842083811759949,
+      "learning_rate": 5.222571070368257e-05,
+      "loss": 0.2893,
+      "step": 529
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3863283395767212,
+      "learning_rate": 5.2072304625964785e-05,
+      "loss": 0.2743,
+      "step": 530
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3817785978317261,
+      "learning_rate": 5.191887900477443e-05,
+      "loss": 0.3419,
+      "step": 531
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3910920023918152,
+      "learning_rate": 5.176543528703657e-05,
+      "loss": 0.3165,
+      "step": 532
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3467884361743927,
+      "learning_rate": 5.161197491984684e-05,
+      "loss": 0.2818,
+      "step": 533
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.42166978120803833,
+      "learning_rate": 5.1458499350458e-05,
+      "loss": 0.2304,
+      "step": 534
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4714958965778351,
+      "learning_rate": 5.130501002626609e-05,
+      "loss": 0.4687,
+      "step": 535
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39768823981285095,
+      "learning_rate": 5.11515083947969e-05,
+      "loss": 0.2814,
+      "step": 536
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3991062641143799,
+      "learning_rate": 5.099799590369231e-05,
+      "loss": 0.3234,
+      "step": 537
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.42261770367622375,
+      "learning_rate": 5.084447400069655e-05,
+      "loss": 0.3404,
+      "step": 538
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.37454599142074585,
+      "learning_rate": 5.069094413364272e-05,
+      "loss": 0.283,
+      "step": 539
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.43171244859695435,
+      "learning_rate": 5.053740775043892e-05,
+      "loss": 0.341,
+      "step": 540
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.46509385108947754,
+      "learning_rate": 5.038386629905475e-05,
+      "loss": 0.3496,
+      "step": 541
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4495503306388855,
+      "learning_rate": 5.023032122750759e-05,
+      "loss": 0.3413,
+      "step": 542
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3662084639072418,
+      "learning_rate": 5.007677398384901e-05,
+      "loss": 0.2604,
+      "step": 543
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40452709794044495,
+      "learning_rate": 4.9923226016151e-05,
+      "loss": 0.3195,
+      "step": 544
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.39293763041496277,
+      "learning_rate": 4.976967877249242e-05,
+      "loss": 0.345,
+      "step": 545
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2965549826622009,
+      "learning_rate": 4.961613370094527e-05,
+      "loss": 0.24,
+      "step": 546
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40824049711227417,
+      "learning_rate": 4.9462592249561095e-05,
+      "loss": 0.3529,
+      "step": 547
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3798341751098633,
+      "learning_rate": 4.9309055866357295e-05,
+      "loss": 0.2365,
+      "step": 548
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40656527876853943,
+      "learning_rate": 4.915552599930345e-05,
+      "loss": 0.341,
+      "step": 549
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3874281644821167,
+      "learning_rate": 4.900200409630771e-05,
+      "loss": 0.2748,
+      "step": 550
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3719668686389923,
+      "learning_rate": 4.884849160520311e-05,
+      "loss": 0.2382,
+      "step": 551
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40866148471832275,
+      "learning_rate": 4.869498997373393e-05,
+      "loss": 0.2957,
+      "step": 552
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4563745856285095,
+      "learning_rate": 4.8541500649542014e-05,
+      "loss": 0.3925,
+      "step": 553
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.41175273060798645,
+      "learning_rate": 4.838802508015316e-05,
+      "loss": 0.2969,
+      "step": 554
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4290417730808258,
+      "learning_rate": 4.823456471296344e-05,
+      "loss": 0.2948,
+      "step": 555
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3427780270576477,
+      "learning_rate": 4.8081120995225574e-05,
+      "loss": 0.2658,
+      "step": 556
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4479462206363678,
+      "learning_rate": 4.792769537403523e-05,
+      "loss": 0.4044,
+      "step": 557
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4526216983795166,
+      "learning_rate": 4.777428929631743e-05,
+      "loss": 0.3701,
+      "step": 558
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.44982680678367615,
+      "learning_rate": 4.762090420881289e-05,
+      "loss": 0.3749,
+      "step": 559
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5037185549736023,
+      "learning_rate": 4.746754155806437e-05,
+      "loss": 0.444,
+      "step": 560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3569648563861847,
+      "learning_rate": 4.731420279040303e-05,
+      "loss": 0.3011,
+      "step": 561
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3476647138595581,
+      "learning_rate": 4.716088935193479e-05,
+      "loss": 0.184,
+      "step": 562
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.45400330424308777,
+      "learning_rate": 4.7007602688526695e-05,
+      "loss": 0.3372,
+      "step": 563
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3246419131755829,
+      "learning_rate": 4.6854344245793305e-05,
+      "loss": 0.2103,
+      "step": 564
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3641456365585327,
+      "learning_rate": 4.6701115469083e-05,
+      "loss": 0.2576,
+      "step": 565
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3781353235244751,
+      "learning_rate": 4.65479178034644e-05,
+      "loss": 0.1963,
+      "step": 566
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.39636749029159546,
+      "learning_rate": 4.6394752693712734e-05,
+      "loss": 0.2883,
+      "step": 567
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.42649081349372864,
+      "learning_rate": 4.624162158429619e-05,
+      "loss": 0.2613,
+      "step": 568
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.48188841342926025,
+      "learning_rate": 4.608852591936231e-05,
+      "loss": 0.3689,
+      "step": 569
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.43533340096473694,
+      "learning_rate": 4.593546714272438e-05,
+      "loss": 0.3374,
+      "step": 570
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.46363407373428345,
+      "learning_rate": 4.578244669784777e-05,
+      "loss": 0.3518,
+      "step": 571
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3086138069629669,
+      "learning_rate": 4.562946602783636e-05,
+      "loss": 0.1972,
+      "step": 572
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5021273493766785,
+      "learning_rate": 4.547652657541897e-05,
+      "loss": 0.4458,
+      "step": 573
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.41480910778045654,
+      "learning_rate": 4.532362978293564e-05,
+      "loss": 0.3708,
+      "step": 574
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3622719943523407,
+      "learning_rate": 4.517077709232411e-05,
+      "loss": 0.2372,
+      "step": 575
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.47259828448295593,
+      "learning_rate": 4.501796994510622e-05,
+      "loss": 0.3861,
+      "step": 576
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2684907913208008,
+      "learning_rate": 4.4865209782374304e-05,
+      "loss": 0.1869,
+      "step": 577
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3525902032852173,
+      "learning_rate": 4.471249804477758e-05,
+      "loss": 0.2118,
+      "step": 578
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.38558393716812134,
+      "learning_rate": 4.4559836172508565e-05,
+      "loss": 0.3187,
+      "step": 579
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4227086901664734,
+      "learning_rate": 4.4407225605289547e-05,
+      "loss": 0.2317,
+      "step": 580
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.32298290729522705,
+      "learning_rate": 4.4254667782358924e-05,
+      "loss": 0.2495,
+      "step": 581
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4255707561969757,
+      "learning_rate": 4.410216414245771e-05,
+      "loss": 0.3593,
+      "step": 582
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.38757482171058655,
+      "learning_rate": 4.394971612381591e-05,
+      "loss": 0.2948,
+      "step": 583
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.41997554898262024,
+      "learning_rate": 4.379732516413897e-05,
+      "loss": 0.2764,
+      "step": 584
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3937298357486725,
+      "learning_rate": 4.364499270059424e-05,
+      "loss": 0.3041,
+      "step": 585
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4118345081806183,
+      "learning_rate": 4.34927201697974e-05,
+      "loss": 0.3201,
+      "step": 586
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38910767436027527,
+      "learning_rate": 4.334050900779893e-05,
+      "loss": 0.2781,
+      "step": 587
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4760296642780304,
+      "learning_rate": 4.3188360650070525e-05,
+      "loss": 0.3535,
+      "step": 588
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.35542789101600647,
+      "learning_rate": 4.303627653149159e-05,
+      "loss": 0.2188,
+      "step": 589
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.37541383504867554,
+      "learning_rate": 4.288425808633575e-05,
+      "loss": 0.2418,
+      "step": 590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4025011360645294,
+      "learning_rate": 4.2732306748257224e-05,
+      "loss": 0.2835,
+      "step": 591
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.35506004095077515,
+      "learning_rate": 4.258042395027738e-05,
+      "loss": 0.2262,
+      "step": 592
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4477950930595398,
+      "learning_rate": 4.2428611124771184e-05,
+      "loss": 0.2769,
+      "step": 593
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3834127187728882,
+      "learning_rate": 4.2276869703453724e-05,
+      "loss": 0.2988,
+      "step": 594
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3999894857406616,
+      "learning_rate": 4.21252011173667e-05,
+      "loss": 0.2643,
+      "step": 595
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3994347155094147,
+      "learning_rate": 4.1973606796864884e-05,
+      "loss": 0.2703,
+      "step": 596
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4385293126106262,
+      "learning_rate": 4.182208817160269e-05,
+      "loss": 0.3637,
+      "step": 597
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3608461618423462,
+      "learning_rate": 4.167064667052065e-05,
+      "loss": 0.2466,
+      "step": 598
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4134450852870941,
+      "learning_rate": 4.151928372183198e-05,
+      "loss": 0.3056,
+      "step": 599
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.41939777135849,
+      "learning_rate": 4.136800075300906e-05,
+      "loss": 0.305,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.2909187376499176,
+      "eval_runtime": 370.6833,
+      "eval_samples_per_second": 0.92,
+      "eval_steps_per_second": 0.232,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4696617126464844,
+      "learning_rate": 4.1216799190770006e-05,
+      "loss": 0.4357,
+      "step": 601
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.33662790060043335,
+      "learning_rate": 4.1065680461065194e-05,
+      "loss": 0.2382,
+      "step": 602
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.38448449969291687,
+      "learning_rate": 4.0914645989063845e-05,
+      "loss": 0.2851,
+      "step": 603
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42738062143325806,
+      "learning_rate": 4.076369719914055e-05,
+      "loss": 0.282,
+      "step": 604
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5082021951675415,
+      "learning_rate": 4.061283551486185e-05,
+      "loss": 0.4564,
+      "step": 605
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4584023356437683,
+      "learning_rate": 4.0462062358972796e-05,
+      "loss": 0.4064,
+      "step": 606
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4898790121078491,
+      "learning_rate": 4.03113791533836e-05,
+      "loss": 0.4274,
+      "step": 607
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4135204255580902,
+      "learning_rate": 4.016078731915608e-05,
+      "loss": 0.2821,
+      "step": 608
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.40426310896873474,
+      "learning_rate": 4.001028827649046e-05,
+      "loss": 0.3121,
+      "step": 609
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3676113188266754,
+      "learning_rate": 3.9859883444711794e-05,
+      "loss": 0.2654,
+      "step": 610
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3547706604003906,
+      "learning_rate": 3.970957424225666e-05,
+      "loss": 0.2522,
+      "step": 611
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.35183659195899963,
+      "learning_rate": 3.955936208665979e-05,
+      "loss": 0.2394,
+      "step": 612
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38255560398101807,
+      "learning_rate": 3.940924839454066e-05,
+      "loss": 0.2843,
+      "step": 613
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.42490723729133606,
+      "learning_rate": 3.925923458159023e-05,
+      "loss": 0.3819,
+      "step": 614
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38167932629585266,
+      "learning_rate": 3.9109322062557424e-05,
+      "loss": 0.28,
+      "step": 615
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.28583797812461853,
+      "learning_rate": 3.895951225123595e-05,
+      "loss": 0.193,
+      "step": 616
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38812220096588135,
+      "learning_rate": 3.880980656045087e-05,
+      "loss": 0.3015,
+      "step": 617
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3436203598976135,
+      "learning_rate": 3.866020640204533e-05,
+      "loss": 0.2642,
+      "step": 618
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3685041069984436,
+      "learning_rate": 3.851071318686721e-05,
+      "loss": 0.2613,
+      "step": 619
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3393908739089966,
+      "learning_rate": 3.8361328324755825e-05,
+      "loss": 0.2301,
+      "step": 620
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.41123929619789124,
+      "learning_rate": 3.821205322452863e-05,
+      "loss": 0.3423,
+      "step": 621
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.42916449904441833,
+      "learning_rate": 3.8062889293967976e-05,
+      "loss": 0.4181,
+      "step": 622
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4151654839515686,
+      "learning_rate": 3.791383793980776e-05,
+      "loss": 0.3556,
+      "step": 623
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.30906325578689575,
+      "learning_rate": 3.7764900567720196e-05,
+      "loss": 0.204,
+      "step": 624
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4097180664539337,
+      "learning_rate": 3.761607858230257e-05,
+      "loss": 0.3965,
+      "step": 625
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3622243404388428,
+      "learning_rate": 3.746737338706397e-05,
+      "loss": 0.2482,
+      "step": 626
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4075562357902527,
+      "learning_rate": 3.7318786384412074e-05,
+      "loss": 0.3814,
+      "step": 627
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3845020830631256,
+      "learning_rate": 3.71703189756399e-05,
+      "loss": 0.3221,
+      "step": 628
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.363413006067276,
+      "learning_rate": 3.70219725609126e-05,
+      "loss": 0.2748,
+      "step": 629
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.38519206643104553,
+      "learning_rate": 3.6873748539254246e-05,
+      "loss": 0.2652,
+      "step": 630
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.43028563261032104,
+      "learning_rate": 3.6725648308534696e-05,
+      "loss": 0.3886,
+      "step": 631
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.39313027262687683,
+      "learning_rate": 3.65776732654563e-05,
+      "loss": 0.3381,
+      "step": 632
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3943227529525757,
+      "learning_rate": 3.6429824805540815e-05,
+      "loss": 0.353,
+      "step": 633
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3451743423938751,
+      "learning_rate": 3.628210432311621e-05,
+      "loss": 0.2154,
+      "step": 634
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4362015426158905,
+      "learning_rate": 3.6134513211303556e-05,
+      "loss": 0.3729,
+      "step": 635
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4508109986782074,
+      "learning_rate": 3.598705286200382e-05,
+      "loss": 0.2969,
+      "step": 636
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4250777065753937,
+      "learning_rate": 3.58397246658848e-05,
+      "loss": 0.3727,
+      "step": 637
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4589483141899109,
+      "learning_rate": 3.5692530012367955e-05,
+      "loss": 0.3555,
+      "step": 638
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.40335288643836975,
+      "learning_rate": 3.554547028961537e-05,
+      "loss": 0.2955,
+      "step": 639
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3560872972011566,
+      "learning_rate": 3.5398546884516604e-05,
+      "loss": 0.2191,
+      "step": 640
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4064195156097412,
+      "learning_rate": 3.5251761182675625e-05,
+      "loss": 0.3488,
+      "step": 641
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4351102113723755,
+      "learning_rate": 3.510511456839777e-05,
+      "loss": 0.3163,
+      "step": 642
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4562118649482727,
+      "learning_rate": 3.495860842467664e-05,
+      "loss": 0.3474,
+      "step": 643
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3723532259464264,
+      "learning_rate": 3.481224413318114e-05,
+      "loss": 0.309,
+      "step": 644
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.43877795338630676,
+      "learning_rate": 3.4666023074242356e-05,
+      "loss": 0.3403,
+      "step": 645
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4374098479747772,
+      "learning_rate": 3.4519946626840573e-05,
+      "loss": 0.3251,
+      "step": 646
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.40907058119773865,
+      "learning_rate": 3.4374016168592296e-05,
+      "loss": 0.3034,
+      "step": 647
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4455272853374481,
+      "learning_rate": 3.422823307573722e-05,
+      "loss": 0.3495,
+      "step": 648
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.424968421459198,
+      "learning_rate": 3.40825987231253e-05,
+      "loss": 0.3816,
+      "step": 649
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.43559131026268005,
+      "learning_rate": 3.393711448420372e-05,
+      "loss": 0.3007,
+      "step": 650
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4118243455886841,
+      "learning_rate": 3.379178173100396e-05,
+      "loss": 0.2909,
+      "step": 651
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.34587720036506653,
+      "learning_rate": 3.364660183412892e-05,
+      "loss": 0.1927,
+      "step": 652
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.46043986082077026,
+      "learning_rate": 3.3501576162739904e-05,
+      "loss": 0.2862,
+      "step": 653
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.398618221282959,
+      "learning_rate": 3.335670608454376e-05,
+      "loss": 0.3402,
+      "step": 654
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3652955889701843,
+      "learning_rate": 3.321199296577998e-05,
+      "loss": 0.2076,
+      "step": 655
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.42596182227134705,
+      "learning_rate": 3.3067438171207766e-05,
+      "loss": 0.318,
+      "step": 656
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.418782502412796,
+      "learning_rate": 3.292304306409325e-05,
+      "loss": 0.3067,
+      "step": 657
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3922344446182251,
+      "learning_rate": 3.2778809006196564e-05,
+      "loss": 0.3039,
+      "step": 658
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.32975706458091736,
+      "learning_rate": 3.263473735775899e-05,
+      "loss": 0.1872,
+      "step": 659
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.377684623003006,
+      "learning_rate": 3.249082947749019e-05,
+      "loss": 0.2447,
+      "step": 660
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4084266722202301,
+      "learning_rate": 3.234708672255538e-05,
+      "loss": 0.2901,
+      "step": 661
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.37359848618507385,
+      "learning_rate": 3.220351044856247e-05,
+      "loss": 0.2602,
+      "step": 662
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3893719017505646,
+      "learning_rate": 3.206010200954935e-05,
+      "loss": 0.2691,
+      "step": 663
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.46557319164276123,
+      "learning_rate": 3.191686275797107e-05,
+      "loss": 0.3768,
+      "step": 664
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3931865692138672,
+      "learning_rate": 3.177379404468715e-05,
+      "loss": 0.2871,
+      "step": 665
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3162769079208374,
+      "learning_rate": 3.163089721894876e-05,
+      "loss": 0.1855,
+      "step": 666
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.40585705637931824,
+      "learning_rate": 3.148817362838606e-05,
+      "loss": 0.3189,
+      "step": 667
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.42030271887779236,
+      "learning_rate": 3.134562461899545e-05,
+      "loss": 0.3164,
+      "step": 668
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.38813257217407227,
+      "learning_rate": 3.120325153512687e-05,
+      "loss": 0.2582,
+      "step": 669
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.40977761149406433,
+      "learning_rate": 3.10610557194712e-05,
+      "loss": 0.2843,
+      "step": 670
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4487515687942505,
+      "learning_rate": 3.091903851304751e-05,
+      "loss": 0.4135,
+      "step": 671
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.34490662813186646,
+      "learning_rate": 3.077720125519042e-05,
+      "loss": 0.2585,
+      "step": 672
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.43439093232154846,
+      "learning_rate": 3.063554528353752e-05,
+      "loss": 0.3276,
+      "step": 673
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3760249614715576,
+      "learning_rate": 3.0494071934016737e-05,
+      "loss": 0.2494,
+      "step": 674
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.40423309803009033,
+      "learning_rate": 3.03527825408337e-05,
+      "loss": 0.3372,
+      "step": 675
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3621990382671356,
+      "learning_rate": 3.0211678436459212e-05,
+      "loss": 0.2584,
+      "step": 676
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.39742204546928406,
+      "learning_rate": 3.0070760951616618e-05,
+      "loss": 0.3147,
+      "step": 677
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4059773087501526,
+      "learning_rate": 2.993003141526933e-05,
+      "loss": 0.3243,
+      "step": 678
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.33496880531311035,
+      "learning_rate": 2.978949115460824e-05,
+      "loss": 0.2166,
+      "step": 679
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4161543548107147,
+      "learning_rate": 2.9649141495039223e-05,
+      "loss": 0.2754,
+      "step": 680
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.389289915561676,
+      "learning_rate": 2.950898376017064e-05,
+      "loss": 0.2811,
+      "step": 681
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3543347418308258,
+      "learning_rate": 2.9369019271800823e-05,
+      "loss": 0.2391,
+      "step": 682
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4240150451660156,
+      "learning_rate": 2.9229249349905684e-05,
+      "loss": 0.3513,
+      "step": 683
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.36569496989250183,
+      "learning_rate": 2.908967531262618e-05,
+      "loss": 0.2202,
+      "step": 684
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.407739520072937,
+      "learning_rate": 2.895029847625595e-05,
+      "loss": 0.2969,
+      "step": 685
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.49572688341140747,
+      "learning_rate": 2.8811120155228844e-05,
+      "loss": 0.4283,
+      "step": 686
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4357757866382599,
+      "learning_rate": 2.8672141662106578e-05,
+      "loss": 0.3349,
+      "step": 687
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.42112380266189575,
+      "learning_rate": 2.853336430756631e-05,
+      "loss": 0.3478,
+      "step": 688
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.48643845319747925,
+      "learning_rate": 2.8394789400388328e-05,
+      "loss": 0.3748,
+      "step": 689
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4737779200077057,
+      "learning_rate": 2.8256418247443662e-05,
+      "loss": 0.3014,
+      "step": 690
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3621935248374939,
+      "learning_rate": 2.811825215368179e-05,
+      "loss": 0.2574,
+      "step": 691
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.438456267118454,
+      "learning_rate": 2.798029242211828e-05,
+      "loss": 0.3644,
+      "step": 692
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4204363226890564,
+      "learning_rate": 2.7842540353822632e-05,
+      "loss": 0.2931,
+      "step": 693
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.366852730512619,
+      "learning_rate": 2.770499724790584e-05,
+      "loss": 0.2649,
+      "step": 694
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.40051698684692383,
+      "learning_rate": 2.7567664401508225e-05,
+      "loss": 0.3275,
+      "step": 695
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4522467255592346,
+      "learning_rate": 2.7430543109787222e-05,
+      "loss": 0.3791,
+      "step": 696
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3605802357196808,
+      "learning_rate": 2.729363466590511e-05,
+      "loss": 0.223,
+      "step": 697
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.38871005177497864,
+      "learning_rate": 2.7156940361016864e-05,
+      "loss": 0.2777,
+      "step": 698
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.39866647124290466,
+      "learning_rate": 2.702046148425795e-05,
+      "loss": 0.2708,
+      "step": 699
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4567882716655731,
+      "learning_rate": 2.688419932273219e-05,
+      "loss": 0.3699,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.2869235873222351,
+      "eval_runtime": 356.7368,
+      "eval_samples_per_second": 0.956,
+      "eval_steps_per_second": 0.241,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.43119126558303833,
+      "learning_rate": 2.6748155161499567e-05,
+      "loss": 0.3449,
+      "step": 701
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.33371806144714355,
+      "learning_rate": 2.6612330283564223e-05,
+      "loss": 0.207,
+      "step": 702
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.36269062757492065,
+      "learning_rate": 2.6476725969862227e-05,
+      "loss": 0.1828,
+      "step": 703
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.346821665763855,
+      "learning_rate": 2.634134349924956e-05,
+      "loss": 0.2371,
+      "step": 704
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4228645861148834,
+      "learning_rate": 2.6206184148490066e-05,
+      "loss": 0.3104,
+      "step": 705
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4131583869457245,
+      "learning_rate": 2.6071249192243363e-05,
+      "loss": 0.3123,
+      "step": 706
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4459822475910187,
+      "learning_rate": 2.5936539903052892e-05,
+      "loss": 0.3429,
+      "step": 707
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4393295645713806,
+      "learning_rate": 2.580205755133384e-05,
+      "loss": 0.3968,
+      "step": 708
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4421556293964386,
+      "learning_rate": 2.5667803405361213e-05,
+      "loss": 0.394,
+      "step": 709
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4462069869041443,
+      "learning_rate": 2.5533778731257824e-05,
+      "loss": 0.3889,
+      "step": 710
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.39122533798217773,
+      "learning_rate": 2.5399984792982457e-05,
+      "loss": 0.2771,
+      "step": 711
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4093492925167084,
+      "learning_rate": 2.5266422852317795e-05,
+      "loss": 0.2972,
+      "step": 712
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4245493412017822,
+      "learning_rate": 2.513309416885865e-05,
+      "loss": 0.3547,
+      "step": 713
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3989643156528473,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 0.2571,
+      "step": 714
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4158961772918701,
+      "learning_rate": 2.4867141600925214e-05,
+      "loss": 0.2883,
+      "step": 715
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.42499738931655884,
+      "learning_rate": 2.4734520224594093e-05,
+      "loss": 0.3401,
+      "step": 716
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4280025362968445,
+      "learning_rate": 2.4602137121731195e-05,
+      "loss": 0.3182,
+      "step": 717
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.398598313331604,
+      "learning_rate": 2.44699935408139e-05,
+      "loss": 0.3046,
+      "step": 718
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2740604877471924,
+      "learning_rate": 2.433809072806081e-05,
+      "loss": 0.1469,
+      "step": 719
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.377769410610199,
+      "learning_rate": 2.4206429927419794e-05,
+      "loss": 0.266,
+      "step": 720
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4317697286605835,
+      "learning_rate": 2.40750123805564e-05,
+      "loss": 0.2638,
+      "step": 721
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4296222925186157,
+      "learning_rate": 2.3943839326842092e-05,
+      "loss": 0.3114,
+      "step": 722
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.35907113552093506,
+      "learning_rate": 2.3812912003342568e-05,
+      "loss": 0.2557,
+      "step": 723
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.44292595982551575,
+      "learning_rate": 2.3682231644806108e-05,
+      "loss": 0.3381,
+      "step": 724
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37591424584388733,
+      "learning_rate": 2.3551799483651894e-05,
+      "loss": 0.2844,
+      "step": 725
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4966159164905548,
+      "learning_rate": 2.342161674995843e-05,
+      "loss": 0.3795,
+      "step": 726
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3967081904411316,
+      "learning_rate": 2.3291684671451903e-05,
+      "loss": 0.3068,
+      "step": 727
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3964744806289673,
+      "learning_rate": 2.3162004473494657e-05,
+      "loss": 0.2539,
+      "step": 728
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37085771560668945,
+      "learning_rate": 2.3032577379073577e-05,
+      "loss": 0.3037,
+      "step": 729
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3845426142215729,
+      "learning_rate": 2.2903404608788582e-05,
+      "loss": 0.2452,
+      "step": 730
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3970223665237427,
+      "learning_rate": 2.2774487380841115e-05,
+      "loss": 0.3195,
+      "step": 731
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4328741431236267,
+      "learning_rate": 2.2645826911022655e-05,
+      "loss": 0.3048,
+      "step": 732
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4075857400894165,
+      "learning_rate": 2.2517424412703252e-05,
+      "loss": 0.2747,
+      "step": 733
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.34755656123161316,
+      "learning_rate": 2.2389281096820075e-05,
+      "loss": 0.2513,
+      "step": 734
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.42572513222694397,
+      "learning_rate": 2.2261398171865978e-05,
+      "loss": 0.3169,
+      "step": 735
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37507641315460205,
+      "learning_rate": 2.2133776843878186e-05,
+      "loss": 0.2619,
+      "step": 736
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4553358554840088,
+      "learning_rate": 2.2006418316426775e-05,
+      "loss": 0.2837,
+      "step": 737
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3961850106716156,
+      "learning_rate": 2.187932379060348e-05,
+      "loss": 0.3218,
+      "step": 738
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3695942163467407,
+      "learning_rate": 2.1752494465010242e-05,
+      "loss": 0.2168,
+      "step": 739
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.40850862860679626,
+      "learning_rate": 2.1625931535747964e-05,
+      "loss": 0.3253,
+      "step": 740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.42420390248298645,
+      "learning_rate": 2.1499636196405228e-05,
+      "loss": 0.3562,
+      "step": 741
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.42372387647628784,
+      "learning_rate": 2.1373609638047032e-05,
+      "loss": 0.2754,
+      "step": 742
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4101789593696594,
+      "learning_rate": 2.1247853049203543e-05,
+      "loss": 0.2667,
+      "step": 743
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4173932671546936,
+      "learning_rate": 2.112236761585892e-05,
+      "loss": 0.3233,
+      "step": 744
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4181278944015503,
+      "learning_rate": 2.09971545214401e-05,
+      "loss": 0.3058,
+      "step": 745
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4128434658050537,
+      "learning_rate": 2.087221494680563e-05,
+      "loss": 0.3331,
+      "step": 746
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.31319257616996765,
+      "learning_rate": 2.074755007023461e-05,
+      "loss": 0.2241,
+      "step": 747
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2864420413970947,
+      "learning_rate": 2.0623161067415463e-05,
+      "loss": 0.158,
+      "step": 748
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4145500957965851,
+      "learning_rate": 2.049904911143492e-05,
+      "loss": 0.3065,
+      "step": 749
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3872825503349304,
+      "learning_rate": 2.0375215372766944e-05,
+      "loss": 0.2999,
+      "step": 750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.41519448161125183,
+      "learning_rate": 2.025166101926168e-05,
+      "loss": 0.4111,
+      "step": 751
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3255583345890045,
+      "learning_rate": 2.012838721613447e-05,
+      "loss": 0.2153,
+      "step": 752
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4237464368343353,
+      "learning_rate": 2.0005395125954812e-05,
+      "loss": 0.3882,
+      "step": 753
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.34982621669769287,
+      "learning_rate": 1.9882685908635462e-05,
+      "loss": 0.2684,
+      "step": 754
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.40125173330307007,
+      "learning_rate": 1.9760260721421426e-05,
+      "loss": 0.3117,
+      "step": 755
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4472535252571106,
+      "learning_rate": 1.963812071887913e-05,
+      "loss": 0.3585,
+      "step": 756
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3477060794830322,
+      "learning_rate": 1.951626705288544e-05,
+      "loss": 0.2959,
+      "step": 757
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3668115735054016,
+      "learning_rate": 1.9394700872616855e-05,
+      "loss": 0.2314,
+      "step": 758
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.40832754969596863,
+      "learning_rate": 1.9273423324538658e-05,
+      "loss": 0.3506,
+      "step": 759
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.36876723170280457,
+      "learning_rate": 1.9152435552394104e-05,
+      "loss": 0.2607,
+      "step": 760
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.33012062311172485,
+      "learning_rate": 1.903173869719362e-05,
+      "loss": 0.2353,
+      "step": 761
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.34388744831085205,
+      "learning_rate": 1.891133389720407e-05,
+      "loss": 0.2664,
+      "step": 762
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4011666476726532,
+      "learning_rate": 1.8791222287937982e-05,
+      "loss": 0.2984,
+      "step": 763
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3983471393585205,
+      "learning_rate": 1.8671405002142918e-05,
+      "loss": 0.2103,
+      "step": 764
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38422495126724243,
+      "learning_rate": 1.855188316979068e-05,
+      "loss": 0.3036,
+      "step": 765
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.43290191888809204,
+      "learning_rate": 1.8432657918066732e-05,
+      "loss": 0.3024,
+      "step": 766
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.41110435128211975,
+      "learning_rate": 1.831373037135955e-05,
+      "loss": 0.3741,
+      "step": 767
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3634113073348999,
+      "learning_rate": 1.8195101651250017e-05,
+      "loss": 0.2117,
+      "step": 768
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5607237219810486,
+      "learning_rate": 1.807677287650083e-05,
+      "loss": 0.2817,
+      "step": 769
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3922412097454071,
+      "learning_rate": 1.7958745163045986e-05,
+      "loss": 0.2926,
+      "step": 770
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.48022595047950745,
+      "learning_rate": 1.7841019623980217e-05,
+      "loss": 0.3129,
+      "step": 771
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.34253695607185364,
+      "learning_rate": 1.77235973695485e-05,
+      "loss": 0.2137,
+      "step": 772
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4008970558643341,
+      "learning_rate": 1.760647950713566e-05,
+      "loss": 0.3095,
+      "step": 773
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3795667290687561,
+      "learning_rate": 1.74896671412558e-05,
+      "loss": 0.2541,
+      "step": 774
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4423181116580963,
+      "learning_rate": 1.737316137354197e-05,
+      "loss": 0.3799,
+      "step": 775
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.39112627506256104,
+      "learning_rate": 1.725696330273575e-05,
+      "loss": 0.3118,
+      "step": 776
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3624914884567261,
+      "learning_rate": 1.7141074024676912e-05,
+      "loss": 0.2802,
+      "step": 777
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3379821181297302,
+      "learning_rate": 1.7025494632293048e-05,
+      "loss": 0.1877,
+      "step": 778
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41077300906181335,
+      "learning_rate": 1.6910226215589303e-05,
+      "loss": 0.3482,
+      "step": 779
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4243071377277374,
+      "learning_rate": 1.6795269861638042e-05,
+      "loss": 0.3369,
+      "step": 780
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3844909965991974,
+      "learning_rate": 1.6680626654568686e-05,
+      "loss": 0.2616,
+      "step": 781
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4580596089363098,
+      "learning_rate": 1.656629767555739e-05,
+      "loss": 0.3754,
+      "step": 782
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4154808521270752,
+      "learning_rate": 1.645228400281689e-05,
+      "loss": 0.2824,
+      "step": 783
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3872333765029907,
+      "learning_rate": 1.6338586711586356e-05,
+      "loss": 0.2446,
+      "step": 784
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.37114283442497253,
+      "learning_rate": 1.6225206874121218e-05,
+      "loss": 0.3106,
+      "step": 785
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.36863473057746887,
+      "learning_rate": 1.6112145559683055e-05,
+      "loss": 0.2771,
+      "step": 786
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4310016930103302,
+      "learning_rate": 1.5999403834529548e-05,
+      "loss": 0.3892,
+      "step": 787
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.36439386010169983,
+      "learning_rate": 1.5886982761904377e-05,
+      "loss": 0.2522,
+      "step": 788
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.42694202065467834,
+      "learning_rate": 1.5774883402027208e-05,
+      "loss": 0.319,
+      "step": 789
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4599052369594574,
+      "learning_rate": 1.5663106812083743e-05,
+      "loss": 0.309,
+      "step": 790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.35446932911872864,
+      "learning_rate": 1.555165404621567e-05,
+      "loss": 0.2609,
+      "step": 791
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.41695091128349304,
+      "learning_rate": 1.5440526155510763e-05,
+      "loss": 0.3687,
+      "step": 792
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4401051998138428,
+      "learning_rate": 1.532972418799298e-05,
+      "loss": 0.3645,
+      "step": 793
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4249185025691986,
+      "learning_rate": 1.5219249188612556e-05,
+      "loss": 0.2889,
+      "step": 794
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4146534204483032,
+      "learning_rate": 1.5109102199236152e-05,
+      "loss": 0.326,
+      "step": 795
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.498243510723114,
+      "learning_rate": 1.4999284258637053e-05,
+      "loss": 0.3464,
+      "step": 796
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.40616828203201294,
+      "learning_rate": 1.488979640248534e-05,
+      "loss": 0.2918,
+      "step": 797
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3539421856403351,
+      "learning_rate": 1.4780639663338125e-05,
+      "loss": 0.2555,
+      "step": 798
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3602314889431,
+      "learning_rate": 1.4671815070629868e-05,
+      "loss": 0.2222,
+      "step": 799
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3262251317501068,
+      "learning_rate": 1.4563323650662586e-05,
+      "loss": 0.1956,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.2838141620159149,
+      "eval_runtime": 358.7052,
+      "eval_samples_per_second": 0.951,
+      "eval_steps_per_second": 0.24,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3623528778553009,
+      "learning_rate": 1.4455166426596223e-05,
+      "loss": 0.2101,
+      "step": 801
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4872315227985382,
+      "learning_rate": 1.434734441843899e-05,
+      "loss": 0.4752,
+      "step": 802
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3752274513244629,
+      "learning_rate": 1.4239858643037751e-05,
+      "loss": 0.2907,
+      "step": 803
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.37255585193634033,
+      "learning_rate": 1.4132710114068426e-05,
+      "loss": 0.2632,
+      "step": 804
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3614441752433777,
+      "learning_rate": 1.4025899842026442e-05,
+      "loss": 0.1775,
+      "step": 805
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3831045925617218,
+      "learning_rate": 1.3919428834217163e-05,
+      "loss": 0.2165,
+      "step": 806
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3734067976474762,
+      "learning_rate": 1.3813298094746491e-05,
+      "loss": 0.2342,
+      "step": 807
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.34909823536872864,
+      "learning_rate": 1.3707508624511262e-05,
+      "loss": 0.2051,
+      "step": 808
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4448155164718628,
+      "learning_rate": 1.36020614211899e-05,
+      "loss": 0.3564,
+      "step": 809
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3687616288661957,
+      "learning_rate": 1.349695747923298e-05,
+      "loss": 0.2529,
+      "step": 810
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3716150224208832,
+      "learning_rate": 1.339219778985385e-05,
+      "loss": 0.2745,
+      "step": 811
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.39968132972717285,
+      "learning_rate": 1.3287783341019278e-05,
+      "loss": 0.2795,
+      "step": 812
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4128952622413635,
+      "learning_rate": 1.3183715117440142e-05,
+      "loss": 0.3301,
+      "step": 813
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.446071058511734,
+      "learning_rate": 1.3079994100562159e-05,
+      "loss": 0.3135,
+      "step": 814
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.40662825107574463,
+      "learning_rate": 1.2976621268556571e-05,
+      "loss": 0.3218,
+      "step": 815
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3803369104862213,
+      "learning_rate": 1.2873597596311027e-05,
+      "loss": 0.2781,
+      "step": 816
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.46125173568725586,
+      "learning_rate": 1.2770924055420258e-05,
+      "loss": 0.2979,
+      "step": 817
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4188249409198761,
+      "learning_rate": 1.2668601614177017e-05,
+      "loss": 0.344,
+      "step": 818
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3977917730808258,
+      "learning_rate": 1.2566631237562893e-05,
+      "loss": 0.2534,
+      "step": 819
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.37687748670578003,
+      "learning_rate": 1.246501388723923e-05,
+      "loss": 0.2566,
+      "step": 820
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.385997474193573,
+      "learning_rate": 1.2363750521538064e-05,
+      "loss": 0.2621,
+      "step": 821
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4117814004421234,
+      "learning_rate": 1.2262842095453065e-05,
+      "loss": 0.3778,
+      "step": 822
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3946126103401184,
+      "learning_rate": 1.2162289560630524e-05,
+      "loss": 0.2904,
+      "step": 823
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4623841941356659,
+      "learning_rate": 1.2062093865360458e-05,
+      "loss": 0.3532,
+      "step": 824
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.36618906259536743,
+      "learning_rate": 1.1962255954567537e-05,
+      "loss": 0.2398,
+      "step": 825
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3738173246383667,
+      "learning_rate": 1.1862776769802276e-05,
+      "loss": 0.2701,
+      "step": 826
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.429414838552475,
+      "learning_rate": 1.1763657249232107e-05,
+      "loss": 0.3365,
+      "step": 827
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.33969846367836,
+      "learning_rate": 1.1664898327632551e-05,
+      "loss": 0.2383,
+      "step": 828
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.41367608308792114,
+      "learning_rate": 1.1566500936378388e-05,
+      "loss": 0.3344,
+      "step": 829
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4863521158695221,
+      "learning_rate": 1.146846600343488e-05,
+      "loss": 0.3525,
+      "step": 830
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4554300308227539,
+      "learning_rate": 1.1370794453349037e-05,
+      "loss": 0.4045,
+      "step": 831
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5504094362258911,
+      "learning_rate": 1.1273487207240847e-05,
+      "loss": 0.2892,
+      "step": 832
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8367794752120972,
+      "learning_rate": 1.1176545182794674e-05,
+      "loss": 0.1812,
+      "step": 833
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.37393510341644287,
+      "learning_rate": 1.1079969294250514e-05,
+      "loss": 0.2687,
+      "step": 834
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4041728973388672,
+      "learning_rate": 1.0983760452395413e-05,
+      "loss": 0.3228,
+      "step": 835
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4162658154964447,
+      "learning_rate": 1.0887919564554894e-05,
+      "loss": 0.3282,
+      "step": 836
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4357545077800751,
+      "learning_rate": 1.079244753458437e-05,
+      "loss": 0.333,
+      "step": 837
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3816227912902832,
+      "learning_rate": 1.0697345262860636e-05,
+      "loss": 0.3241,
+      "step": 838
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3980691730976105,
+      "learning_rate": 1.0602613646273374e-05,
+      "loss": 0.357,
+      "step": 839
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4512961506843567,
+      "learning_rate": 1.0508253578216693e-05,
+      "loss": 0.3677,
+      "step": 840
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4330853521823883,
+      "learning_rate": 1.0414265948580692e-05,
+      "loss": 0.3242,
+      "step": 841
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.40338146686553955,
+      "learning_rate": 1.032065164374313e-05,
+      "loss": 0.2474,
+      "step": 842
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.347898006439209,
+      "learning_rate": 1.0227411546560962e-05,
+      "loss": 0.2574,
+      "step": 843
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3843633532524109,
+      "learning_rate": 1.01345465363621e-05,
+      "loss": 0.2945,
+      "step": 844
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4198579490184784,
+      "learning_rate": 1.0042057488937067e-05,
+      "loss": 0.295,
+      "step": 845
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4400755763053894,
+      "learning_rate": 9.949945276530781e-06,
+      "loss": 0.3899,
+      "step": 846
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3292907178401947,
+      "learning_rate": 9.858210767834292e-06,
+      "loss": 0.2506,
+      "step": 847
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.43633392453193665,
+      "learning_rate": 9.766854827976617e-06,
+      "loss": 0.2899,
+      "step": 848
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3870198130607605,
+      "learning_rate": 9.675878318516546e-06,
+      "loss": 0.2179,
+      "step": 849
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5287765264511108,
+      "learning_rate": 9.58528209743459e-06,
+      "loss": 0.4015,
+      "step": 850
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3869243264198303,
+      "learning_rate": 9.495067019124792e-06,
+      "loss": 0.3133,
+      "step": 851
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.42058876156806946,
+      "learning_rate": 9.405233934386726e-06,
+      "loss": 0.2845,
+      "step": 852
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3767407536506653,
+      "learning_rate": 9.315783690417479e-06,
+      "loss": 0.244,
+      "step": 853
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39176231622695923,
+      "learning_rate": 9.226717130803636e-06,
+      "loss": 0.2598,
+      "step": 854
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3902047574520111,
+      "learning_rate": 9.138035095513336e-06,
+      "loss": 0.3257,
+      "step": 855
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3235481083393097,
+      "learning_rate": 9.049738420888349e-06,
+      "loss": 0.2546,
+      "step": 856
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39131927490234375,
+      "learning_rate": 8.961827939636196e-06,
+      "loss": 0.2679,
+      "step": 857
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3007175624370575,
+      "learning_rate": 8.87430448082227e-06,
+      "loss": 0.1367,
+      "step": 858
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3458656668663025,
+      "learning_rate": 8.787168869862066e-06,
+      "loss": 0.2285,
+      "step": 859
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.40784284472465515,
+      "learning_rate": 8.700421928513352e-06,
+      "loss": 0.3907,
+      "step": 860
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4105082154273987,
+      "learning_rate": 8.614064474868421e-06,
+      "loss": 0.2758,
+      "step": 861
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4141014516353607,
+      "learning_rate": 8.528097323346407e-06,
+      "loss": 0.3264,
+      "step": 862
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.32178714871406555,
+      "learning_rate": 8.442521284685573e-06,
+      "loss": 0.19,
+      "step": 863
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.35529130697250366,
+      "learning_rate": 8.357337165935675e-06,
+      "loss": 0.2665,
+      "step": 864
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.44634607434272766,
+      "learning_rate": 8.27254577045039e-06,
+      "loss": 0.3463,
+      "step": 865
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37395426630973816,
+      "learning_rate": 8.188147897879667e-06,
+      "loss": 0.2235,
+      "step": 866
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3743934631347656,
+      "learning_rate": 8.104144344162229e-06,
+      "loss": 0.2493,
+      "step": 867
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3435682952404022,
+      "learning_rate": 8.02053590151805e-06,
+      "loss": 0.2583,
+      "step": 868
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4007546603679657,
+      "learning_rate": 7.937323358440935e-06,
+      "loss": 0.265,
+      "step": 869
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5093558430671692,
+      "learning_rate": 7.854507499691005e-06,
+      "loss": 0.1579,
+      "step": 870
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.39399033784866333,
+      "learning_rate": 7.772089106287344e-06,
+      "loss": 0.314,
+      "step": 871
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.41897690296173096,
+      "learning_rate": 7.690068955500624e-06,
+      "loss": 0.3568,
+      "step": 872
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3838496804237366,
+      "learning_rate": 7.608447820845771e-06,
+      "loss": 0.2371,
+      "step": 873
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3652224540710449,
+      "learning_rate": 7.527226472074678e-06,
+      "loss": 0.2628,
+      "step": 874
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4027518630027771,
+      "learning_rate": 7.446405675168938e-06,
+      "loss": 0.3228,
+      "step": 875
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.36246979236602783,
+      "learning_rate": 7.365986192332625e-06,
+      "loss": 0.1864,
+      "step": 876
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.36685219407081604,
+      "learning_rate": 7.285968781985092e-06,
+      "loss": 0.2548,
+      "step": 877
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3361806869506836,
+      "learning_rate": 7.206354198753862e-06,
+      "loss": 0.1993,
+      "step": 878
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.43565836548805237,
+      "learning_rate": 7.127143193467445e-06,
+      "loss": 0.3497,
+      "step": 879
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5036295056343079,
+      "learning_rate": 7.048336513148307e-06,
+      "loss": 0.3685,
+      "step": 880
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5267510414123535,
+      "learning_rate": 6.969934901005809e-06,
+      "loss": 0.4194,
+      "step": 881
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3988897204399109,
+      "learning_rate": 6.8919390964291895e-06,
+      "loss": 0.3389,
+      "step": 882
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4020831882953644,
+      "learning_rate": 6.814349834980621e-06,
+      "loss": 0.2958,
+      "step": 883
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4476851522922516,
+      "learning_rate": 6.7371678483882264e-06,
+      "loss": 0.4031,
+      "step": 884
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3566264808177948,
+      "learning_rate": 6.660393864539222e-06,
+      "loss": 0.258,
+      "step": 885
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.34493035078048706,
+      "learning_rate": 6.58402860747302e-06,
+      "loss": 0.2073,
+      "step": 886
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3956291377544403,
+      "learning_rate": 6.508072797374454e-06,
+      "loss": 0.3207,
+      "step": 887
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.45581164956092834,
+      "learning_rate": 6.4325271505669025e-06,
+      "loss": 0.3234,
+      "step": 888
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4293762147426605,
+      "learning_rate": 6.3573923795056e-06,
+      "loss": 0.3209,
+      "step": 889
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4189924895763397,
+      "learning_rate": 6.282669192770896e-06,
+      "loss": 0.3334,
+      "step": 890
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.42287614941596985,
+      "learning_rate": 6.208358295061572e-06,
+      "loss": 0.2984,
+      "step": 891
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3633008897304535,
+      "learning_rate": 6.134460387188206e-06,
+      "loss": 0.2521,
+      "step": 892
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.38413792848587036,
+      "learning_rate": 6.060976166066546e-06,
+      "loss": 0.2638,
+      "step": 893
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.443373441696167,
+      "learning_rate": 5.9879063247109405e-06,
+      "loss": 0.3292,
+      "step": 894
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.41159743070602417,
+      "learning_rate": 5.915251552227829e-06,
+      "loss": 0.2839,
+      "step": 895
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4896499812602997,
+      "learning_rate": 5.8430125338092115e-06,
+      "loss": 0.386,
+      "step": 896
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3715815842151642,
+      "learning_rate": 5.771189950726191e-06,
+      "loss": 0.2282,
+      "step": 897
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4447186291217804,
+      "learning_rate": 5.699784480322568e-06,
+      "loss": 0.3621,
+      "step": 898
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4121861755847931,
+      "learning_rate": 5.628796796008434e-06,
+      "loss": 0.323,
+      "step": 899
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3753719925880432,
+      "learning_rate": 5.558227567253832e-06,
+      "loss": 0.28,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.279915988445282,
+      "eval_runtime": 369.8376,
+      "eval_samples_per_second": 0.922,
+      "eval_steps_per_second": 0.233,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.44346171617507935,
+      "learning_rate": 5.488077459582425e-06,
+      "loss": 0.3365,
+      "step": 901
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3690701425075531,
+      "learning_rate": 5.418347134565249e-06,
+      "loss": 0.2932,
+      "step": 902
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3719370365142822,
+      "learning_rate": 5.349037249814442e-06,
+      "loss": 0.2902,
+      "step": 903
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4262860119342804,
+      "learning_rate": 5.28014845897708e-06,
+      "loss": 0.2894,
+      "step": 904
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.42489689588546753,
+      "learning_rate": 5.211681411728969e-06,
+      "loss": 0.3145,
+      "step": 905
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3699342906475067,
+      "learning_rate": 5.143636753768549e-06,
+      "loss": 0.2345,
+      "step": 906
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.36667463183403015,
+      "learning_rate": 5.076015126810784e-06,
+      "loss": 0.2978,
+      "step": 907
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38998737931251526,
+      "learning_rate": 5.008817168581137e-06,
+      "loss": 0.2756,
+      "step": 908
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4218217730522156,
+      "learning_rate": 4.94204351280953e-06,
+      "loss": 0.3131,
+      "step": 909
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4226126968860626,
+      "learning_rate": 4.8756947892243725e-06,
+      "loss": 0.3172,
+      "step": 910
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.37969115376472473,
+      "learning_rate": 4.809771623546627e-06,
+      "loss": 0.218,
+      "step": 911
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3295484185218811,
+      "learning_rate": 4.744274637483936e-06,
+      "loss": 0.2061,
+      "step": 912
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3455501198768616,
+      "learning_rate": 4.6792044487247e-06,
+      "loss": 0.2871,
+      "step": 913
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3930600881576538,
+      "learning_rate": 4.614561670932288e-06,
+      "loss": 0.2772,
+      "step": 914
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3832755386829376,
+      "learning_rate": 4.550346913739256e-06,
+      "loss": 0.2336,
+      "step": 915
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3684132993221283,
+      "learning_rate": 4.486560782741578e-06,
+      "loss": 0.2225,
+      "step": 916
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3281688392162323,
+      "learning_rate": 4.423203879492943e-06,
+      "loss": 0.2159,
+      "step": 917
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4367421865463257,
+      "learning_rate": 4.360276801499086e-06,
+      "loss": 0.3513,
+      "step": 918
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.44129833579063416,
+      "learning_rate": 4.29778014221216e-06,
+      "loss": 0.2732,
+      "step": 919
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.43012675642967224,
+      "learning_rate": 4.2357144910251e-06,
+      "loss": 0.3074,
+      "step": 920
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.45049425959587097,
+      "learning_rate": 4.174080433266136e-06,
+      "loss": 0.3341,
+      "step": 921
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.45134514570236206,
+      "learning_rate": 4.112878550193194e-06,
+      "loss": 0.3232,
+      "step": 922
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3490968942642212,
+      "learning_rate": 4.05210941898847e-06,
+      "loss": 0.2193,
+      "step": 923
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3718993365764618,
+      "learning_rate": 3.991773612752952e-06,
+      "loss": 0.2418,
+      "step": 924
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4038708508014679,
+      "learning_rate": 3.9318717005010495e-06,
+      "loss": 0.3119,
+      "step": 925
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3851058781147003,
+      "learning_rate": 3.872404247155193e-06,
+      "loss": 0.3203,
+      "step": 926
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.42264580726623535,
+      "learning_rate": 3.8133718135405285e-06,
+      "loss": 0.3537,
+      "step": 927
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4049663543701172,
+      "learning_rate": 3.754774956379614e-06,
+      "loss": 0.2828,
+      "step": 928
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.43319347500801086,
+      "learning_rate": 3.696614228287187e-06,
+      "loss": 0.3109,
+      "step": 929
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4413582980632782,
+      "learning_rate": 3.6388901777649477e-06,
+      "loss": 0.3179,
+      "step": 930
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.44206950068473816,
+      "learning_rate": 3.581603349196372e-06,
+      "loss": 0.2914,
+      "step": 931
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4736064374446869,
+      "learning_rate": 3.5247542828415747e-06,
+      "loss": 0.393,
+      "step": 932
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.36942192912101746,
+      "learning_rate": 3.468343514832251e-06,
+      "loss": 0.2324,
+      "step": 933
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3748895823955536,
+      "learning_rate": 3.4123715771665786e-06,
+      "loss": 0.2102,
+      "step": 934
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4034070074558258,
+      "learning_rate": 3.356838997704226e-06,
+      "loss": 0.3282,
+      "step": 935
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3851265013217926,
+      "learning_rate": 3.3017463001613626e-06,
+      "loss": 0.2537,
+      "step": 936
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3298608958721161,
+      "learning_rate": 3.2470940041057106e-06,
+      "loss": 0.2088,
+      "step": 937
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3934982419013977,
+      "learning_rate": 3.1928826249516987e-06,
+      "loss": 0.2113,
+      "step": 938
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.42488792538642883,
+      "learning_rate": 3.139112673955513e-06,
+      "loss": 0.2801,
+      "step": 939
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4066559672355652,
+      "learning_rate": 3.08578465821035e-06,
+      "loss": 0.3073,
+      "step": 940
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3734951615333557,
+      "learning_rate": 3.0328990806415934e-06,
+      "loss": 0.2836,
+      "step": 941
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.47129473090171814,
+      "learning_rate": 2.9804564400020994e-06,
+      "loss": 0.3448,
+      "step": 942
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41345497965812683,
+      "learning_rate": 2.9284572308674596e-06,
+      "loss": 0.3359,
+      "step": 943
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4121784269809723,
+      "learning_rate": 2.8769019436313715e-06,
+      "loss": 0.3181,
+      "step": 944
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3603481352329254,
+      "learning_rate": 2.825791064500993e-06,
+      "loss": 0.2822,
+      "step": 945
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4569970667362213,
+      "learning_rate": 2.7751250754923573e-06,
+      "loss": 0.3064,
+      "step": 946
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3882192373275757,
+      "learning_rate": 2.7249044544258363e-06,
+      "loss": 0.3168,
+      "step": 947
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4478338658809662,
+      "learning_rate": 2.675129674921639e-06,
+      "loss": 0.4022,
+      "step": 948
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5042359232902527,
+      "learning_rate": 2.625801206395312e-06,
+      "loss": 0.391,
+      "step": 949
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.39275994896888733,
+      "learning_rate": 2.576919514053355e-06,
+      "loss": 0.2659,
+      "step": 950
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4786848723888397,
+      "learning_rate": 2.528485058888813e-06,
+      "loss": 0.3756,
+      "step": 951
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.40453627705574036,
+      "learning_rate": 2.48049829767692e-06,
+      "loss": 0.3275,
+      "step": 952
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.49131420254707336,
+      "learning_rate": 2.4329596829708144e-06,
+      "loss": 0.4037,
+      "step": 953
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.37989288568496704,
+      "learning_rate": 2.385869663097251e-06,
+      "loss": 0.2858,
+      "step": 954
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.44397759437561035,
+      "learning_rate": 2.3392286821523724e-06,
+      "loss": 0.3501,
+      "step": 955
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3832888901233673,
+      "learning_rate": 2.2930371799975594e-06,
+      "loss": 0.2205,
+      "step": 956
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3776226341724396,
+      "learning_rate": 2.2472955922552163e-06,
+      "loss": 0.2503,
+      "step": 957
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.42176002264022827,
+      "learning_rate": 2.202004350304715e-06,
+      "loss": 0.3537,
+      "step": 958
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.41694408655166626,
+      "learning_rate": 2.157163881278312e-06,
+      "loss": 0.3156,
+      "step": 959
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.39139553904533386,
+      "learning_rate": 2.112774608057111e-06,
+      "loss": 0.2938,
+      "step": 960
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.43866458535194397,
+      "learning_rate": 2.068836949267089e-06,
+      "loss": 0.3437,
+      "step": 961
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4038626253604889,
+      "learning_rate": 2.0253513192751373e-06,
+      "loss": 0.2769,
+      "step": 962
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3881865441799164,
+      "learning_rate": 1.982318128185151e-06,
+      "loss": 0.2959,
+      "step": 963
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.42929449677467346,
+      "learning_rate": 1.9397377818341944e-06,
+      "loss": 0.3153,
+      "step": 964
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.43852919340133667,
+      "learning_rate": 1.8976106817886196e-06,
+      "loss": 0.3636,
+      "step": 965
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.47231245040893555,
+      "learning_rate": 1.855937225340315e-06,
+      "loss": 0.3046,
+      "step": 966
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3930777907371521,
+      "learning_rate": 1.8147178055029579e-06,
+      "loss": 0.2753,
+      "step": 967
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.334532767534256,
+      "learning_rate": 1.7739528110083004e-06,
+      "loss": 0.2482,
+      "step": 968
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.41035714745521545,
+      "learning_rate": 1.7336426263024897e-06,
+      "loss": 0.3114,
+      "step": 969
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.31711286306381226,
+      "learning_rate": 1.6937876315424705e-06,
+      "loss": 0.2026,
+      "step": 970
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3743811547756195,
+      "learning_rate": 1.6543882025923886e-06,
+      "loss": 0.2697,
+      "step": 971
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.43929824233055115,
+      "learning_rate": 1.6154447110200255e-06,
+      "loss": 0.3436,
+      "step": 972
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3947059214115143,
+      "learning_rate": 1.5769575240933422e-06,
+      "loss": 0.2185,
+      "step": 973
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5468947291374207,
+      "learning_rate": 1.5389270047769578e-06,
+      "loss": 0.4863,
+      "step": 974
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3948152959346771,
+      "learning_rate": 1.5013535117287647e-06,
+      "loss": 0.3023,
+      "step": 975
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.35228264331817627,
+      "learning_rate": 1.4642373992965364e-06,
+      "loss": 0.2013,
+      "step": 976
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3721064329147339,
+      "learning_rate": 1.4275790175145908e-06,
+      "loss": 0.2285,
+      "step": 977
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3736189305782318,
+      "learning_rate": 1.3913787121004719e-06,
+      "loss": 0.2567,
+      "step": 978
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4247582256793976,
+      "learning_rate": 1.3556368244517114e-06,
+      "loss": 0.3116,
+      "step": 979
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3985781967639923,
+      "learning_rate": 1.3203536916425841e-06,
+      "loss": 0.2872,
+      "step": 980
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4070413112640381,
+      "learning_rate": 1.2855296464209688e-06,
+      "loss": 0.3075,
+      "step": 981
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.456635445356369,
+      "learning_rate": 1.2511650172051637e-06,
+      "loss": 0.4196,
+      "step": 982
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3921426236629486,
+      "learning_rate": 1.217260128080816e-06,
+      "loss": 0.292,
+      "step": 983
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.37394100427627563,
+      "learning_rate": 1.1838152987978578e-06,
+      "loss": 0.2828,
+      "step": 984
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.42904219031333923,
+      "learning_rate": 1.1508308447674975e-06,
+      "loss": 0.3018,
+      "step": 985
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4142332375049591,
+      "learning_rate": 1.1183070770592441e-06,
+      "loss": 0.2906,
+      "step": 986
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3668891191482544,
+      "learning_rate": 1.0862443023979651e-06,
+      "loss": 0.2611,
+      "step": 987
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32188835740089417,
+      "learning_rate": 1.0546428231609895e-06,
+      "loss": 0.2276,
+      "step": 988
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4997420310974121,
+      "learning_rate": 1.0235029373752758e-06,
+      "loss": 0.3959,
+      "step": 989
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.39162763953208923,
+      "learning_rate": 9.928249387145982e-07,
+      "loss": 0.2642,
+      "step": 990
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3797048330307007,
+      "learning_rate": 9.6260911649676e-07,
+      "loss": 0.2132,
+      "step": 991
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3422527015209198,
+      "learning_rate": 9.32855755680867e-07,
+      "loss": 0.2203,
+      "step": 992
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3924298584461212,
+      "learning_rate": 9.035651368646648e-07,
+      "loss": 0.3098,
+      "step": 993
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.48407474160194397,
+      "learning_rate": 8.747375362818666e-07,
+      "loss": 0.4102,
+      "step": 994
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4377146065235138,
+      "learning_rate": 8.463732257995571e-07,
+      "loss": 0.3217,
+      "step": 995
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.31259143352508545,
+      "learning_rate": 8.184724729156379e-07,
+      "loss": 0.209,
+      "step": 996
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.42829185724258423,
+      "learning_rate": 7.910355407562742e-07,
+      "loss": 0.2971,
+      "step": 997
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.38974568247795105,
+      "learning_rate": 7.640626880734581e-07,
+      "loss": 0.2204,
+      "step": 998
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.31262338161468506,
+      "learning_rate": 7.375541692425326e-07,
+      "loss": 0.2123,
+      "step": 999
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.43367284536361694,
+      "learning_rate": 7.115102342598101e-07,
+      "loss": 0.34,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.27866458892822266,
+      "eval_runtime": 357.1324,
+      "eval_samples_per_second": 0.955,
+      "eval_steps_per_second": 0.241,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4019174575805664,
+      "learning_rate": 6.859311287402081e-07,
+      "loss": 0.254,
+      "step": 1001
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.33488696813583374,
+      "learning_rate": 6.608170939149283e-07,
+      "loss": 0.1977,
+      "step": 1002
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3944094777107239,
+      "learning_rate": 6.361683666291973e-07,
+      "loss": 0.3229,
+      "step": 1003
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.46557244658470154,
+      "learning_rate": 6.119851793400189e-07,
+      "loss": 0.31,
+      "step": 1004
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4170675575733185,
+      "learning_rate": 5.882677601139919e-07,
+      "loss": 0.2635,
+      "step": 1005
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4197085201740265,
+      "learning_rate": 5.650163326251345e-07,
+      "loss": 0.2622,
+      "step": 1006
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4764816462993622,
+      "learning_rate": 5.422311161528193e-07,
+      "loss": 0.3789,
+      "step": 1007
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.42266231775283813,
+      "learning_rate": 5.199123255796634e-07,
+      "loss": 0.2654,
+      "step": 1008
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3167967200279236,
+      "learning_rate": 4.980601713895305e-07,
+      "loss": 0.1765,
+      "step": 1009
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.40985655784606934,
+      "learning_rate": 4.7667485966552685e-07,
+      "loss": 0.2456,
+      "step": 1010
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38095444440841675,
+      "learning_rate": 4.55756592088058e-07,
+      "loss": 0.3334,
+      "step": 1011
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4163905084133148,
+      "learning_rate": 4.353055659329419e-07,
+      "loss": 0.378,
+      "step": 1012
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.45960038900375366,
+      "learning_rate": 4.153219740695435e-07,
+      "loss": 0.3774,
+      "step": 1013
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.48693540692329407,
+      "learning_rate": 3.958060049589485e-07,
+      "loss": 0.3586,
+      "step": 1014
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3643259108066559,
+      "learning_rate": 3.767578426521923e-07,
+      "loss": 0.2477,
+      "step": 1015
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.36645421385765076,
+      "learning_rate": 3.5817766678850615e-07,
+      "loss": 0.2515,
+      "step": 1016
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3555854260921478,
+      "learning_rate": 3.4006565259366295e-07,
+      "loss": 0.254,
+      "step": 1017
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.38294434547424316,
+      "learning_rate": 3.224219708782894e-07,
+      "loss": 0.2859,
+      "step": 1018
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4626649022102356,
+      "learning_rate": 3.052467880362675e-07,
+      "loss": 0.3928,
+      "step": 1019
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3951588273048401,
+      "learning_rate": 2.88540266043158e-07,
+      "loss": 0.3058,
+      "step": 1020
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.42776766419410706,
+      "learning_rate": 2.723025624546849e-07,
+      "loss": 0.3367,
+      "step": 1021
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.37951818108558655,
+      "learning_rate": 2.5653383040524227e-07,
+      "loss": 0.3122,
+      "step": 1022
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.41030153632164,
+      "learning_rate": 2.412342186064565e-07,
+      "loss": 0.236,
+      "step": 1023
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4126613140106201,
+      "learning_rate": 2.2640387134577058e-07,
+      "loss": 0.3693,
+      "step": 1024
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3438760042190552,
+      "learning_rate": 2.1204292848509555e-07,
+      "loss": 0.2377,
+      "step": 1025
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4316646158695221,
+      "learning_rate": 1.98151525459489e-07,
+      "loss": 0.2699,
+      "step": 1026
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3581903576850891,
+      "learning_rate": 1.8472979327587292e-07,
+      "loss": 0.2106,
+      "step": 1027
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.43876421451568604,
+      "learning_rate": 1.717778585118013e-07,
+      "loss": 0.3461,
+      "step": 1028
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3924015760421753,
+      "learning_rate": 1.5929584331427216e-07,
+      "loss": 0.3041,
+      "step": 1029
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.49308621883392334,
+      "learning_rate": 1.4728386539856754e-07,
+      "loss": 0.3544,
+      "step": 1030
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4344156086444855,
+      "learning_rate": 1.3574203804713747e-07,
+      "loss": 0.4094,
+      "step": 1031
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.39321210980415344,
+      "learning_rate": 1.246704701085566e-07,
+      "loss": 0.2796,
+      "step": 1032
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3914424479007721,
+      "learning_rate": 1.1406926599646373e-07,
+      "loss": 0.2821,
+      "step": 1033
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.36496034264564514,
+      "learning_rate": 1.0393852568860719e-07,
+      "loss": 0.233,
+      "step": 1034
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2860527038574219,
+      "learning_rate": 9.427834472588992e-08,
+      "loss": 0.1483,
+      "step": 1035
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4255082607269287,
+      "learning_rate": 8.508881421145365e-08,
+      "loss": 0.2685,
+      "step": 1036
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3710958659648895,
+      "learning_rate": 7.637002080985168e-08,
+      "loss": 0.2375,
+      "step": 1037
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.41288140416145325,
+      "learning_rate": 6.812204674619959e-08,
+      "loss": 0.3242,
+      "step": 1038
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3869653642177582,
+      "learning_rate": 6.034496980542037e-08,
+      "loss": 0.2814,
+      "step": 1039
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39398565888404846,
+      "learning_rate": 5.303886333151153e-08,
+      "loss": 0.2723,
+      "step": 1040
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3731752038002014,
+      "learning_rate": 4.620379622682358e-08,
+      "loss": 0.2714,
+      "step": 1041
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.36814767122268677,
+      "learning_rate": 3.9839832951465986e-08,
+      "loss": 0.2471,
+      "step": 1042
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3668307960033417,
+      "learning_rate": 3.394703352263551e-08,
+      "loss": 0.2658,
+      "step": 1043
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4195619523525238,
+      "learning_rate": 2.8525453514099965e-08,
+      "loss": 0.3402,
+      "step": 1044
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3522312641143799,
+      "learning_rate": 2.3575144055643094e-08,
+      "loss": 0.1953,
+      "step": 1045
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3813425898551941,
+      "learning_rate": 1.9096151832609375e-08,
+      "loss": 0.2474,
+      "step": 1046
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3777915835380554,
+      "learning_rate": 1.5088519085437734e-08,
+      "loss": 0.3004,
+      "step": 1047
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3522752523422241,
+      "learning_rate": 1.1552283609272962e-08,
+      "loss": 0.2251,
+      "step": 1048
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.383852481842041,
+      "learning_rate": 8.487478753615997e-09,
+      "loss": 0.3306,
+      "step": 1049
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4403495490550995,
+      "learning_rate": 5.894133422001957e-09,
+      "loss": 0.3644,
+      "step": 1050
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3327406346797943,
+      "learning_rate": 3.772272071722593e-09,
+      "loss": 0.2278,
+      "step": 1051
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3965804874897003,
+      "learning_rate": 2.1219147136264382e-09,
+      "loss": 0.2637,
+      "step": 1052
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4187352955341339,
+      "learning_rate": 9.43076911874563e-10,
+      "loss": 0.3538,
+      "step": 1053
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.43565842509269714,
+      "learning_rate": 2.3576978384065584e-10,
+      "loss": 0.319,
+      "step": 1054
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.35923850536346436,
+      "learning_rate": 0.0,
+      "loss": 0.2605,
+      "step": 1055
+    },
+    {
+      "epoch": 1.0,
+      "step": 1055,
+      "total_flos": 1.1584683941193318e+17,
+      "train_loss": 0.312048421468215,
+      "train_runtime": 97657.3561,
+      "train_samples_per_second": 0.346,
+      "train_steps_per_second": 0.011
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1055,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 1.1584683941193318e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}