diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/README.md b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/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_long_lora_agg_fc4_s2048/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_long_lora_agg_fc4_s2048/adapter_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b52c78046f475459e6bcbd293ac11b130e0465a6
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/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": [
+    "o_proj",
+    "v_proj",
+    "up_proj",
+    "q_proj",
+    "down_proj",
+    "gate_proj",
+    "k_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_long_lora_agg_fc4_s2048/adapter_model.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..3962575f5a30a8714d642063a00ae111ba780254
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ac49bbeffef9a922b0c552724bd8e8e56947977f9f1e712332b8f35b71ee962
+size 692127130
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..a48d68518cce4598c62f52102a6c0a9e1adc9faa
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/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": 2048,
+  "mm_streaming_patches_per_frame": 729,
+  "mm_streaming_state_dim": 1152,
+  "mm_streaming_vision_chunk_size": 8,
+  "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_long_lora_agg_fc4_s2048/generation_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/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_long_lora_agg_fc4_s2048/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_long_lora_agg_fc4_s2048/non_lora_trainables.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..af94ac00203e2ef28ca549b275800bd309119244
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:962ae8d5423d00e9a671fbcb830867cfbb404d5998696725f4238dcc4340a805
+size 211441788
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/trainer_state.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..b0fed2eea4634add8de0bbcec45da549d4500e8a
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-7B-Instruct-nextqa_shuffled-streaming_baseline_long_lora_agg_fc4_s2048/trainer_state.json
@@ -0,0 +1,14968 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9992893521260215,
+  "eval_steps": 100,
+  "global_step": 2110,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 15.237822532653809,
+      "learning_rate": 1.5625e-06,
+      "loss": 1.1209,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.978849411010742,
+      "learning_rate": 3.125e-06,
+      "loss": 1.0615,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.950687408447266,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 1.1521,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.257516860961914,
+      "learning_rate": 6.25e-06,
+      "loss": 1.153,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 9.888971328735352,
+      "learning_rate": 7.8125e-06,
+      "loss": 0.9857,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.9900736808776855,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.9075,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.3253068923950195,
+      "learning_rate": 1.09375e-05,
+      "loss": 0.706,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.3243162631988525,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5837,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9421330690383911,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 0.4717,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.7493945360183716,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.4322,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2339285612106323,
+      "learning_rate": 1.71875e-05,
+      "loss": 0.3638,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9665519595146179,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.3097,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9759274125099182,
+      "learning_rate": 2.0312500000000002e-05,
+      "loss": 0.3883,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9763781428337097,
+      "learning_rate": 2.1875e-05,
+      "loss": 0.394,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9802628755569458,
+      "learning_rate": 2.34375e-05,
+      "loss": 0.5679,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.9744460582733154,
+      "learning_rate": 2.5e-05,
+      "loss": 0.4158,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.638671875,
+      "learning_rate": 2.6562500000000002e-05,
+      "loss": 0.46,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6141244769096375,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 0.3058,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5645075440406799,
+      "learning_rate": 2.96875e-05,
+      "loss": 0.3685,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5531061887741089,
+      "learning_rate": 3.125e-05,
+      "loss": 0.2694,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.594106137752533,
+      "learning_rate": 3.2812500000000005e-05,
+      "loss": 0.3911,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.43871158361434937,
+      "learning_rate": 3.4375e-05,
+      "loss": 0.1808,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5581795573234558,
+      "learning_rate": 3.59375e-05,
+      "loss": 0.3056,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6783905029296875,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.4777,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5139743685722351,
+      "learning_rate": 3.90625e-05,
+      "loss": 0.2325,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5659346580505371,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.3626,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6135446429252625,
+      "learning_rate": 4.21875e-05,
+      "loss": 0.3643,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.583995521068573,
+      "learning_rate": 4.375e-05,
+      "loss": 0.3376,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5542269349098206,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 0.3419,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5978631973266602,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.3109,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5763736963272095,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 0.314,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7249500751495361,
+      "learning_rate": 5e-05,
+      "loss": 0.3297,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6381571292877197,
+      "learning_rate": 5.15625e-05,
+      "loss": 0.3615,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5638049244880676,
+      "learning_rate": 5.3125000000000004e-05,
+      "loss": 0.3711,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7142037749290466,
+      "learning_rate": 5.46875e-05,
+      "loss": 0.4801,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5071200728416443,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 0.2957,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5474107265472412,
+      "learning_rate": 5.78125e-05,
+      "loss": 0.3743,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6407603025436401,
+      "learning_rate": 5.9375e-05,
+      "loss": 0.3761,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5440143942832947,
+      "learning_rate": 6.0937500000000004e-05,
+      "loss": 0.2903,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4683564007282257,
+      "learning_rate": 6.25e-05,
+      "loss": 0.3108,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6603513956069946,
+      "learning_rate": 6.40625e-05,
+      "loss": 0.4439,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5716879963874817,
+      "learning_rate": 6.562500000000001e-05,
+      "loss": 0.3462,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49622324109077454,
+      "learning_rate": 6.71875e-05,
+      "loss": 0.3607,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5135329365730286,
+      "learning_rate": 6.875e-05,
+      "loss": 0.2945,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5732174515724182,
+      "learning_rate": 7.031250000000001e-05,
+      "loss": 0.4966,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4719146490097046,
+      "learning_rate": 7.1875e-05,
+      "loss": 0.2699,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4554480314254761,
+      "learning_rate": 7.34375e-05,
+      "loss": 0.2604,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4951834976673126,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.3965,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.48007503151893616,
+      "learning_rate": 7.65625e-05,
+      "loss": 0.3457,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5186550617218018,
+      "learning_rate": 7.8125e-05,
+      "loss": 0.3244,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5569330453872681,
+      "learning_rate": 7.96875e-05,
+      "loss": 0.3824,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5017390251159668,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 0.3485,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5576614737510681,
+      "learning_rate": 8.28125e-05,
+      "loss": 0.4349,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5364780426025391,
+      "learning_rate": 8.4375e-05,
+      "loss": 0.4398,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49022629857063293,
+      "learning_rate": 8.593750000000001e-05,
+      "loss": 0.3773,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.453789621591568,
+      "learning_rate": 8.75e-05,
+      "loss": 0.2774,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5246427655220032,
+      "learning_rate": 8.90625e-05,
+      "loss": 0.4227,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41480061411857605,
+      "learning_rate": 9.062500000000001e-05,
+      "loss": 0.2433,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5278374552726746,
+      "learning_rate": 9.21875e-05,
+      "loss": 0.4155,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.49553951621055603,
+      "learning_rate": 9.375e-05,
+      "loss": 0.3176,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4902215301990509,
+      "learning_rate": 9.53125e-05,
+      "loss": 0.3941,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48030799627304077,
+      "learning_rate": 9.687500000000001e-05,
+      "loss": 0.3148,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4178484082221985,
+      "learning_rate": 9.84375e-05,
+      "loss": 0.3052,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4596129059791565,
+      "learning_rate": 0.0001,
+      "loss": 0.3398,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5021818280220032,
+      "learning_rate": 9.999994105751931e-05,
+      "loss": 0.445,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4354231059551239,
+      "learning_rate": 9.999976423021617e-05,
+      "loss": 0.2977,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3848458528518677,
+      "learning_rate": 9.999946951850749e-05,
+      "loss": 0.2433,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3888322412967682,
+      "learning_rate": 9.999905692308813e-05,
+      "loss": 0.2128,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4350215792655945,
+      "learning_rate": 9.999852644493086e-05,
+      "loss": 0.3048,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4273955225944519,
+      "learning_rate": 9.999787808528638e-05,
+      "loss": 0.2868,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5307473540306091,
+      "learning_rate": 9.999711184568333e-05,
+      "loss": 0.3551,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.39743614196777344,
+      "learning_rate": 9.999622772792828e-05,
+      "loss": 0.2051,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5220218896865845,
+      "learning_rate": 9.99952257341057e-05,
+      "loss": 0.3899,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4202117919921875,
+      "learning_rate": 9.9994105866578e-05,
+      "loss": 0.2674,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4513757526874542,
+      "learning_rate": 9.99928681279855e-05,
+      "loss": 0.3151,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3918982446193695,
+      "learning_rate": 9.999151252124639e-05,
+      "loss": 0.2964,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4863679111003876,
+      "learning_rate": 9.99900390495568e-05,
+      "loss": 0.3388,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4588763117790222,
+      "learning_rate": 9.998844771639073e-05,
+      "loss": 0.3994,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4668104648590088,
+      "learning_rate": 9.998673852550007e-05,
+      "loss": 0.3329,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4076758325099945,
+      "learning_rate": 9.998491148091457e-05,
+      "loss": 0.2882,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4311077296733856,
+      "learning_rate": 9.998296658694185e-05,
+      "loss": 0.3591,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.446644127368927,
+      "learning_rate": 9.99809038481674e-05,
+      "loss": 0.309,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.49731069803237915,
+      "learning_rate": 9.997872326945452e-05,
+      "loss": 0.3379,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.49003592133522034,
+      "learning_rate": 9.997642485594436e-05,
+      "loss": 0.3922,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.45873504877090454,
+      "learning_rate": 9.997400861305591e-05,
+      "loss": 0.3206,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.47314688563346863,
+      "learning_rate": 9.99714745464859e-05,
+      "loss": 0.3457,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5191733837127686,
+      "learning_rate": 9.996882266220895e-05,
+      "loss": 0.4175,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4319663643836975,
+      "learning_rate": 9.996605296647737e-05,
+      "loss": 0.3258,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4966694116592407,
+      "learning_rate": 9.996316546582128e-05,
+      "loss": 0.3735,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.42036789655685425,
+      "learning_rate": 9.996016016704854e-05,
+      "loss": 0.2943,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.41199979186058044,
+      "learning_rate": 9.995703707724474e-05,
+      "loss": 0.2001,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4653874337673187,
+      "learning_rate": 9.995379620377318e-05,
+      "loss": 0.3629,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4610278904438019,
+      "learning_rate": 9.995043755427487e-05,
+      "loss": 0.4227,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4858045279979706,
+      "learning_rate": 9.99469611366685e-05,
+      "loss": 0.3623,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5327629446983337,
+      "learning_rate": 9.99433669591504e-05,
+      "loss": 0.3934,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5167436599731445,
+      "learning_rate": 9.993965503019457e-05,
+      "loss": 0.2763,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4890263080596924,
+      "learning_rate": 9.993582535855263e-05,
+      "loss": 0.3024,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4444724917411804,
+      "learning_rate": 9.99318779532538e-05,
+      "loss": 0.3149,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5230104923248291,
+      "learning_rate": 9.992781282360486e-05,
+      "loss": 0.3753,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4286307692527771,
+      "learning_rate": 9.992362997919016e-05,
+      "loss": 0.3394,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.31564271450042725,
+      "eval_runtime": 365.9199,
+      "eval_samples_per_second": 0.932,
+      "eval_steps_per_second": 0.235,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4629710912704468,
+      "learning_rate": 9.991932942987158e-05,
+      "loss": 0.2936,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.46175268292427063,
+      "learning_rate": 9.991491118578856e-05,
+      "loss": 0.2935,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.520667552947998,
+      "learning_rate": 9.991037525735794e-05,
+      "loss": 0.3563,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47858309745788574,
+      "learning_rate": 9.990572165527413e-05,
+      "loss": 0.379,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45799320936203003,
+      "learning_rate": 9.990095039050886e-05,
+      "loss": 0.2684,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44723790884017944,
+      "learning_rate": 9.98960614743114e-05,
+      "loss": 0.3224,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.41240695118904114,
+      "learning_rate": 9.989105491820831e-05,
+      "loss": 0.2737,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4039137661457062,
+      "learning_rate": 9.988593073400354e-05,
+      "loss": 0.2722,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.44280770421028137,
+      "learning_rate": 9.98806889337784e-05,
+      "loss": 0.2832,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4407433271408081,
+      "learning_rate": 9.987532952989145e-05,
+      "loss": 0.2301,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.47410380840301514,
+      "learning_rate": 9.986985253497858e-05,
+      "loss": 0.3283,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4414031207561493,
+      "learning_rate": 9.986425796195287e-05,
+      "loss": 0.29,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5073645114898682,
+      "learning_rate": 9.985854582400465e-05,
+      "loss": 0.384,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.554114580154419,
+      "learning_rate": 9.985271613460144e-05,
+      "loss": 0.424,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4527111053466797,
+      "learning_rate": 9.984676890748787e-05,
+      "loss": 0.2739,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4637836217880249,
+      "learning_rate": 9.984070415668574e-05,
+      "loss": 0.2877,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5850847959518433,
+      "learning_rate": 9.983452189649387e-05,
+      "loss": 0.3105,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44936996698379517,
+      "learning_rate": 9.982822214148821e-05,
+      "loss": 0.3222,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.47328370809555054,
+      "learning_rate": 9.982180490652165e-05,
+      "loss": 0.2715,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43314439058303833,
+      "learning_rate": 9.981527020672413e-05,
+      "loss": 0.2613,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.41822177171707153,
+      "learning_rate": 9.98086180575025e-05,
+      "loss": 0.2551,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4786980450153351,
+      "learning_rate": 9.980184847454052e-05,
+      "loss": 0.3385,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5026147365570068,
+      "learning_rate": 9.979496147379883e-05,
+      "loss": 0.4051,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.440973162651062,
+      "learning_rate": 9.978795707151492e-05,
+      "loss": 0.3668,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.41149234771728516,
+      "learning_rate": 9.978083528420303e-05,
+      "loss": 0.271,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.45806679129600525,
+      "learning_rate": 9.977359612865423e-05,
+      "loss": 0.3395,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.46893593668937683,
+      "learning_rate": 9.976623962193626e-05,
+      "loss": 0.3832,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44203895330429077,
+      "learning_rate": 9.975876578139355e-05,
+      "loss": 0.3006,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4383772611618042,
+      "learning_rate": 9.975117462464715e-05,
+      "loss": 0.3487,
+      "step": 129
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5309450626373291,
+      "learning_rate": 9.974346616959476e-05,
+      "loss": 0.4675,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4859173595905304,
+      "learning_rate": 9.973564043441057e-05,
+      "loss": 0.362,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4932904541492462,
+      "learning_rate": 9.972769743754531e-05,
+      "loss": 0.4058,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4158841371536255,
+      "learning_rate": 9.971963719772621e-05,
+      "loss": 0.3053,
+      "step": 133
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4543537199497223,
+      "learning_rate": 9.971145973395684e-05,
+      "loss": 0.3822,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4134504497051239,
+      "learning_rate": 9.970316506551725e-05,
+      "loss": 0.2866,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.40971729159355164,
+      "learning_rate": 9.969475321196373e-05,
+      "loss": 0.328,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.47445958852767944,
+      "learning_rate": 9.968622419312894e-05,
+      "loss": 0.3776,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3023748993873596,
+      "learning_rate": 9.96775780291217e-05,
+      "loss": 0.1728,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.36804643273353577,
+      "learning_rate": 9.96688147403271e-05,
+      "loss": 0.2365,
+      "step": 139
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.48352986574172974,
+      "learning_rate": 9.965993434740633e-05,
+      "loss": 0.3968,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4240727424621582,
+      "learning_rate": 9.965093687129668e-05,
+      "loss": 0.2805,
+      "step": 141
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4886670708656311,
+      "learning_rate": 9.96418223332115e-05,
+      "loss": 0.3706,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.49167630076408386,
+      "learning_rate": 9.963259075464011e-05,
+      "loss": 0.3145,
+      "step": 143
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.49180886149406433,
+      "learning_rate": 9.962324215734782e-05,
+      "loss": 0.3781,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5207163095474243,
+      "learning_rate": 9.961377656337579e-05,
+      "loss": 0.3709,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5093397498130798,
+      "learning_rate": 9.960419399504106e-05,
+      "loss": 0.3376,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4132647216320038,
+      "learning_rate": 9.959449447493643e-05,
+      "loss": 0.2815,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4356156587600708,
+      "learning_rate": 9.958467802593046e-05,
+      "loss": 0.3049,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3599499464035034,
+      "learning_rate": 9.957474467116739e-05,
+      "loss": 0.2231,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.47369691729545593,
+      "learning_rate": 9.956469443406706e-05,
+      "loss": 0.3745,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4666244387626648,
+      "learning_rate": 9.955452733832493e-05,
+      "loss": 0.2914,
+      "step": 151
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4214138984680176,
+      "learning_rate": 9.954424340791196e-05,
+      "loss": 0.3394,
+      "step": 152
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.5188649892807007,
+      "learning_rate": 9.953384266707453e-05,
+      "loss": 0.3629,
+      "step": 153
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.36275699734687805,
+      "learning_rate": 9.952332514033447e-05,
+      "loss": 0.2597,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5535814762115479,
+      "learning_rate": 9.951269085248898e-05,
+      "loss": 0.3767,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39487582445144653,
+      "learning_rate": 9.950193982861047e-05,
+      "loss": 0.3031,
+      "step": 156
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.439854234457016,
+      "learning_rate": 9.949107209404665e-05,
+      "loss": 0.2869,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.43165817856788635,
+      "learning_rate": 9.948008767442034e-05,
+      "loss": 0.3189,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3992789387702942,
+      "learning_rate": 9.946898659562951e-05,
+      "loss": 0.2916,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4712015986442566,
+      "learning_rate": 9.94577688838472e-05,
+      "loss": 0.3313,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3452704846858978,
+      "learning_rate": 9.944643456552133e-05,
+      "loss": 0.2081,
+      "step": 161
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4708247780799866,
+      "learning_rate": 9.943498366737487e-05,
+      "loss": 0.3108,
+      "step": 162
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4105573296546936,
+      "learning_rate": 9.942341621640558e-05,
+      "loss": 0.289,
+      "step": 163
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4130713641643524,
+      "learning_rate": 9.941173223988602e-05,
+      "loss": 0.2653,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4205513000488281,
+      "learning_rate": 9.93999317653635e-05,
+      "loss": 0.3213,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.45442888140678406,
+      "learning_rate": 9.938801482065998e-05,
+      "loss": 0.3155,
+      "step": 166
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.45634108781814575,
+      "learning_rate": 9.937598143387205e-05,
+      "loss": 0.262,
+      "step": 167
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.45718148350715637,
+      "learning_rate": 9.93638316333708e-05,
+      "loss": 0.3475,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.44168248772621155,
+      "learning_rate": 9.935156544780183e-05,
+      "loss": 0.3125,
+      "step": 169
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.30777597427368164,
+      "learning_rate": 9.933918290608509e-05,
+      "loss": 0.2084,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4320216178894043,
+      "learning_rate": 9.932668403741488e-05,
+      "loss": 0.3634,
+      "step": 171
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5343308448791504,
+      "learning_rate": 9.93140688712598e-05,
+      "loss": 0.4555,
+      "step": 172
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3641800284385681,
+      "learning_rate": 9.93013374373626e-05,
+      "loss": 0.2591,
+      "step": 173
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.43759191036224365,
+      "learning_rate": 9.928848976574019e-05,
+      "loss": 0.3146,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.48238664865493774,
+      "learning_rate": 9.927552588668351e-05,
+      "loss": 0.4215,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.46212807297706604,
+      "learning_rate": 9.926244583075748e-05,
+      "loss": 0.3382,
+      "step": 176
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4418286383152008,
+      "learning_rate": 9.924924962880093e-05,
+      "loss": 0.289,
+      "step": 177
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4455692172050476,
+      "learning_rate": 9.923593731192655e-05,
+      "loss": 0.3406,
+      "step": 178
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.48894476890563965,
+      "learning_rate": 9.922250891152077e-05,
+      "loss": 0.3743,
+      "step": 179
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3994240164756775,
+      "learning_rate": 9.920896445924373e-05,
+      "loss": 0.3203,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39229097962379456,
+      "learning_rate": 9.919530398702917e-05,
+      "loss": 0.2568,
+      "step": 181
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.43320316076278687,
+      "learning_rate": 9.918152752708438e-05,
+      "loss": 0.3597,
+      "step": 182
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.42106303572654724,
+      "learning_rate": 9.916763511189009e-05,
+      "loss": 0.3318,
+      "step": 183
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6433641910552979,
+      "learning_rate": 9.915362677420044e-05,
+      "loss": 0.4223,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.438774585723877,
+      "learning_rate": 9.913950254704291e-05,
+      "loss": 0.3317,
+      "step": 185
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.48059192299842834,
+      "learning_rate": 9.912526246371814e-05,
+      "loss": 0.3465,
+      "step": 186
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5154823064804077,
+      "learning_rate": 9.911090655779997e-05,
+      "loss": 0.4461,
+      "step": 187
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.45711418986320496,
+      "learning_rate": 9.909643486313533e-05,
+      "loss": 0.3429,
+      "step": 188
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.40631893277168274,
+      "learning_rate": 9.908184741384412e-05,
+      "loss": 0.2816,
+      "step": 189
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3920607268810272,
+      "learning_rate": 9.906714424431913e-05,
+      "loss": 0.2797,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4251416325569153,
+      "learning_rate": 9.905232538922605e-05,
+      "loss": 0.3328,
+      "step": 191
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4302024841308594,
+      "learning_rate": 9.903739088350325e-05,
+      "loss": 0.2626,
+      "step": 192
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4971807301044464,
+      "learning_rate": 9.902234076236181e-05,
+      "loss": 0.4817,
+      "step": 193
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.42372027039527893,
+      "learning_rate": 9.900717506128541e-05,
+      "loss": 0.3111,
+      "step": 194
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5906328558921814,
+      "learning_rate": 9.899189381603018e-05,
+      "loss": 0.4306,
+      "step": 195
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4526865780353546,
+      "learning_rate": 9.897649706262473e-05,
+      "loss": 0.381,
+      "step": 196
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4398210942745209,
+      "learning_rate": 9.896098483736995e-05,
+      "loss": 0.3377,
+      "step": 197
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.34487393498420715,
+      "learning_rate": 9.894535717683902e-05,
+      "loss": 0.2247,
+      "step": 198
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4142954647541046,
+      "learning_rate": 9.892961411787724e-05,
+      "loss": 0.3525,
+      "step": 199
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3846341073513031,
+      "learning_rate": 9.891375569760204e-05,
+      "loss": 0.3027,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.3086688220500946,
+      "eval_runtime": 370.1341,
+      "eval_samples_per_second": 0.921,
+      "eval_steps_per_second": 0.232,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.40650591254234314,
+      "learning_rate": 9.889778195340278e-05,
+      "loss": 0.3217,
+      "step": 201
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4331025779247284,
+      "learning_rate": 9.888169292294076e-05,
+      "loss": 0.3099,
+      "step": 202
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3728821575641632,
+      "learning_rate": 9.886548864414907e-05,
+      "loss": 0.2645,
+      "step": 203
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.37846335768699646,
+      "learning_rate": 9.884916915523251e-05,
+      "loss": 0.1673,
+      "step": 204
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.41704538464546204,
+      "learning_rate": 9.883273449466754e-05,
+      "loss": 0.3619,
+      "step": 205
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3432096540927887,
+      "learning_rate": 9.881618470120215e-05,
+      "loss": 0.1831,
+      "step": 206
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5096997618675232,
+      "learning_rate": 9.879951981385578e-05,
+      "loss": 0.3516,
+      "step": 207
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5211670994758606,
+      "learning_rate": 9.87827398719192e-05,
+      "loss": 0.3777,
+      "step": 208
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.41217195987701416,
+      "learning_rate": 9.876584491495447e-05,
+      "loss": 0.3066,
+      "step": 209
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.41415494680404663,
+      "learning_rate": 9.874883498279484e-05,
+      "loss": 0.2337,
+      "step": 210
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.5228385925292969,
+      "learning_rate": 9.87317101155446e-05,
+      "loss": 0.4126,
+      "step": 211
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.46517354249954224,
+      "learning_rate": 9.871447035357903e-05,
+      "loss": 0.3502,
+      "step": 212
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38248777389526367,
+      "learning_rate": 9.869711573754433e-05,
+      "loss": 0.2462,
+      "step": 213
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4908299744129181,
+      "learning_rate": 9.867964630835743e-05,
+      "loss": 0.3624,
+      "step": 214
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3816898763179779,
+      "learning_rate": 9.866206210720599e-05,
+      "loss": 0.2318,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3995162844657898,
+      "learning_rate": 9.864436317554831e-05,
+      "loss": 0.3028,
+      "step": 216
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.388855904340744,
+      "learning_rate": 9.862654955511309e-05,
+      "loss": 0.2712,
+      "step": 217
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.48248806595802307,
+      "learning_rate": 9.860862128789953e-05,
+      "loss": 0.3985,
+      "step": 218
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.35596340894699097,
+      "learning_rate": 9.859057841617709e-05,
+      "loss": 0.2317,
+      "step": 219
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3823217451572418,
+      "learning_rate": 9.857242098248542e-05,
+      "loss": 0.3024,
+      "step": 220
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.41867077350616455,
+      "learning_rate": 9.855414902963429e-05,
+      "loss": 0.3312,
+      "step": 221
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3848174214363098,
+      "learning_rate": 9.853576260070347e-05,
+      "loss": 0.2644,
+      "step": 222
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4058029353618622,
+      "learning_rate": 9.851726173904264e-05,
+      "loss": 0.3396,
+      "step": 223
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.41045626997947693,
+      "learning_rate": 9.849864648827125e-05,
+      "loss": 0.3246,
+      "step": 224
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4648635685443878,
+      "learning_rate": 9.847991689227847e-05,
+      "loss": 0.3693,
+      "step": 225
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.43114838004112244,
+      "learning_rate": 9.846107299522304e-05,
+      "loss": 0.2988,
+      "step": 226
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4011891782283783,
+      "learning_rate": 9.844211484153325e-05,
+      "loss": 0.2322,
+      "step": 227
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5050569176673889,
+      "learning_rate": 9.842304247590667e-05,
+      "loss": 0.4172,
+      "step": 228
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.398188978433609,
+      "learning_rate": 9.840385594331022e-05,
+      "loss": 0.2583,
+      "step": 229
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5038655400276184,
+      "learning_rate": 9.838455528897998e-05,
+      "loss": 0.3802,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4851484000682831,
+      "learning_rate": 9.836514055842107e-05,
+      "loss": 0.3421,
+      "step": 231
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.44400739669799805,
+      "learning_rate": 9.834561179740762e-05,
+      "loss": 0.2635,
+      "step": 232
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.46352294087409973,
+      "learning_rate": 9.832596905198254e-05,
+      "loss": 0.3635,
+      "step": 233
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4697592854499817,
+      "learning_rate": 9.830621236845754e-05,
+      "loss": 0.376,
+      "step": 234
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3571726381778717,
+      "learning_rate": 9.828634179341292e-05,
+      "loss": 0.2274,
+      "step": 235
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4293690025806427,
+      "learning_rate": 9.826635737369751e-05,
+      "loss": 0.3484,
+      "step": 236
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4125104248523712,
+      "learning_rate": 9.82462591564286e-05,
+      "loss": 0.3147,
+      "step": 237
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4091227054595947,
+      "learning_rate": 9.82260471889917e-05,
+      "loss": 0.3167,
+      "step": 238
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.481508731842041,
+      "learning_rate": 9.82057215190406e-05,
+      "loss": 0.3415,
+      "step": 239
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3781089782714844,
+      "learning_rate": 9.818528219449705e-05,
+      "loss": 0.2453,
+      "step": 240
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.41487643122673035,
+      "learning_rate": 9.816472926355086e-05,
+      "loss": 0.2593,
+      "step": 241
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4563277065753937,
+      "learning_rate": 9.814406277465968e-05,
+      "loss": 0.3939,
+      "step": 242
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4035716652870178,
+      "learning_rate": 9.812328277654888e-05,
+      "loss": 0.3133,
+      "step": 243
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4190554916858673,
+      "learning_rate": 9.810238931821138e-05,
+      "loss": 0.3386,
+      "step": 244
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3812645673751831,
+      "learning_rate": 9.808138244890775e-05,
+      "loss": 0.2906,
+      "step": 245
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4821646809577942,
+      "learning_rate": 9.806026221816581e-05,
+      "loss": 0.338,
+      "step": 246
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.49207499623298645,
+      "learning_rate": 9.803902867578075e-05,
+      "loss": 0.3934,
+      "step": 247
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.47490590810775757,
+      "learning_rate": 9.801768187181486e-05,
+      "loss": 0.3348,
+      "step": 248
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4753337502479553,
+      "learning_rate": 9.799622185659747e-05,
+      "loss": 0.3751,
+      "step": 249
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4608486294746399,
+      "learning_rate": 9.797464868072488e-05,
+      "loss": 0.3529,
+      "step": 250
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4108763039112091,
+      "learning_rate": 9.795296239506012e-05,
+      "loss": 0.313,
+      "step": 251
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3977440595626831,
+      "learning_rate": 9.793116305073292e-05,
+      "loss": 0.2943,
+      "step": 252
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.5365564823150635,
+      "learning_rate": 9.79092506991396e-05,
+      "loss": 0.4601,
+      "step": 253
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4634103775024414,
+      "learning_rate": 9.788722539194291e-05,
+      "loss": 0.4009,
+      "step": 254
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4129694700241089,
+      "learning_rate": 9.786508718107185e-05,
+      "loss": 0.3239,
+      "step": 255
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.49253442883491516,
+      "learning_rate": 9.784283611872169e-05,
+      "loss": 0.3366,
+      "step": 256
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4075859487056732,
+      "learning_rate": 9.782047225735376e-05,
+      "loss": 0.3109,
+      "step": 257
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.46275150775909424,
+      "learning_rate": 9.77979956496953e-05,
+      "loss": 0.3875,
+      "step": 258
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.48548048734664917,
+      "learning_rate": 9.777540634873938e-05,
+      "loss": 0.4006,
+      "step": 259
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4042569100856781,
+      "learning_rate": 9.77527044077448e-05,
+      "loss": 0.3243,
+      "step": 260
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.39857277274131775,
+      "learning_rate": 9.772988988023588e-05,
+      "loss": 0.3712,
+      "step": 261
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.35716918110847473,
+      "learning_rate": 9.770696282000244e-05,
+      "loss": 0.2549,
+      "step": 262
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3978220820426941,
+      "learning_rate": 9.768392328109958e-05,
+      "loss": 0.3261,
+      "step": 263
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3265039920806885,
+      "learning_rate": 9.766077131784763e-05,
+      "loss": 0.2236,
+      "step": 264
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.41396430134773254,
+      "learning_rate": 9.763750698483191e-05,
+      "loss": 0.2935,
+      "step": 265
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3817710280418396,
+      "learning_rate": 9.761413033690276e-05,
+      "loss": 0.2254,
+      "step": 266
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.48103228211402893,
+      "learning_rate": 9.759064142917526e-05,
+      "loss": 0.3524,
+      "step": 267
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.49803054332733154,
+      "learning_rate": 9.756704031702918e-05,
+      "loss": 0.4147,
+      "step": 268
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5670162439346313,
+      "learning_rate": 9.754332705610889e-05,
+      "loss": 0.4562,
+      "step": 269
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5007535815238953,
+      "learning_rate": 9.751950170232309e-05,
+      "loss": 0.3926,
+      "step": 270
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.46831515431404114,
+      "learning_rate": 9.74955643118448e-05,
+      "loss": 0.3883,
+      "step": 271
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.48109665513038635,
+      "learning_rate": 9.74715149411112e-05,
+      "loss": 0.3139,
+      "step": 272
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4074740409851074,
+      "learning_rate": 9.744735364682346e-05,
+      "loss": 0.2642,
+      "step": 273
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.497220903635025,
+      "learning_rate": 9.742308048594665e-05,
+      "loss": 0.3941,
+      "step": 274
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.39224934577941895,
+      "learning_rate": 9.739869551570959e-05,
+      "loss": 0.2804,
+      "step": 275
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.45215389132499695,
+      "learning_rate": 9.73741987936047e-05,
+      "loss": 0.3263,
+      "step": 276
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.44326817989349365,
+      "learning_rate": 9.734959037738787e-05,
+      "loss": 0.3057,
+      "step": 277
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3992641270160675,
+      "learning_rate": 9.732487032507837e-05,
+      "loss": 0.2643,
+      "step": 278
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5571819543838501,
+      "learning_rate": 9.730003869495863e-05,
+      "loss": 0.3915,
+      "step": 279
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.42717769742012024,
+      "learning_rate": 9.727509554557417e-05,
+      "loss": 0.2602,
+      "step": 280
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4302416443824768,
+      "learning_rate": 9.725004093573342e-05,
+      "loss": 0.2889,
+      "step": 281
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.49743029475212097,
+      "learning_rate": 9.722487492450764e-05,
+      "loss": 0.4062,
+      "step": 282
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.41181784868240356,
+      "learning_rate": 9.719959757123071e-05,
+      "loss": 0.2386,
+      "step": 283
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5087786316871643,
+      "learning_rate": 9.717420893549902e-05,
+      "loss": 0.3246,
+      "step": 284
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.44081348180770874,
+      "learning_rate": 9.714870907717133e-05,
+      "loss": 0.2847,
+      "step": 285
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5425200462341309,
+      "learning_rate": 9.712309805636863e-05,
+      "loss": 0.3557,
+      "step": 286
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3979805111885071,
+      "learning_rate": 9.709737593347403e-05,
+      "loss": 0.2499,
+      "step": 287
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4212340712547302,
+      "learning_rate": 9.707154276913256e-05,
+      "loss": 0.3016,
+      "step": 288
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.43242713809013367,
+      "learning_rate": 9.704559862425102e-05,
+      "loss": 0.3097,
+      "step": 289
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.4071507453918457,
+      "learning_rate": 9.701954355999792e-05,
+      "loss": 0.3136,
+      "step": 290
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.34054338932037354,
+      "learning_rate": 9.699337763780326e-05,
+      "loss": 0.2096,
+      "step": 291
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.42213180661201477,
+      "learning_rate": 9.696710091935841e-05,
+      "loss": 0.2894,
+      "step": 292
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.442929744720459,
+      "learning_rate": 9.694071346661599e-05,
+      "loss": 0.2959,
+      "step": 293
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5444359183311462,
+      "learning_rate": 9.691421534178966e-05,
+      "loss": 0.4137,
+      "step": 294
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4757278263568878,
+      "learning_rate": 9.688760660735402e-05,
+      "loss": 0.3632,
+      "step": 295
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3607676923274994,
+      "learning_rate": 9.686088732604449e-05,
+      "loss": 0.2564,
+      "step": 296
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.42650651931762695,
+      "learning_rate": 9.683405756085708e-05,
+      "loss": 0.3479,
+      "step": 297
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.41149193048477173,
+      "learning_rate": 9.68071173750483e-05,
+      "loss": 0.3,
+      "step": 298
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.40139949321746826,
+      "learning_rate": 9.678006683213503e-05,
+      "loss": 0.2885,
+      "step": 299
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.39154571294784546,
+      "learning_rate": 9.675290599589429e-05,
+      "loss": 0.3254,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.30830395221710205,
+      "eval_runtime": 383.0548,
+      "eval_samples_per_second": 0.89,
+      "eval_steps_per_second": 0.225,
+      "step": 300
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.48314720392227173,
+      "learning_rate": 9.672563493036318e-05,
+      "loss": 0.3904,
+      "step": 301
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.41254061460494995,
+      "learning_rate": 9.669825369983865e-05,
+      "loss": 0.2759,
+      "step": 302
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.43461254239082336,
+      "learning_rate": 9.667076236887743e-05,
+      "loss": 0.3408,
+      "step": 303
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.41292914748191833,
+      "learning_rate": 9.664316100229578e-05,
+      "loss": 0.2703,
+      "step": 304
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4454971253871918,
+      "learning_rate": 9.661544966516945e-05,
+      "loss": 0.3517,
+      "step": 305
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.39257603883743286,
+      "learning_rate": 9.658762842283343e-05,
+      "loss": 0.2951,
+      "step": 306
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5099194645881653,
+      "learning_rate": 9.655969734088184e-05,
+      "loss": 0.4317,
+      "step": 307
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4619141221046448,
+      "learning_rate": 9.653165648516776e-05,
+      "loss": 0.4081,
+      "step": 308
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.41113242506980896,
+      "learning_rate": 9.65035059218031e-05,
+      "loss": 0.3034,
+      "step": 309
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4275076687335968,
+      "learning_rate": 9.647524571715843e-05,
+      "loss": 0.2988,
+      "step": 310
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4362778961658478,
+      "learning_rate": 9.644687593786282e-05,
+      "loss": 0.3385,
+      "step": 311
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4126892387866974,
+      "learning_rate": 9.641839665080363e-05,
+      "loss": 0.3244,
+      "step": 312
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4750313460826874,
+      "learning_rate": 9.638980792312651e-05,
+      "loss": 0.3571,
+      "step": 313
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.45338037610054016,
+      "learning_rate": 9.636110982223505e-05,
+      "loss": 0.3222,
+      "step": 314
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4365720748901367,
+      "learning_rate": 9.633230241579076e-05,
+      "loss": 0.3308,
+      "step": 315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4666551947593689,
+      "learning_rate": 9.630338577171282e-05,
+      "loss": 0.3538,
+      "step": 316
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4274444878101349,
+      "learning_rate": 9.627435995817799e-05,
+      "loss": 0.3255,
+      "step": 317
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.41035231947898865,
+      "learning_rate": 9.624522504362039e-05,
+      "loss": 0.2654,
+      "step": 318
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.44391104578971863,
+      "learning_rate": 9.621598109673142e-05,
+      "loss": 0.3204,
+      "step": 319
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.44411224126815796,
+      "learning_rate": 9.618662818645948e-05,
+      "loss": 0.2686,
+      "step": 320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.49596402049064636,
+      "learning_rate": 9.615716638200993e-05,
+      "loss": 0.3106,
+      "step": 321
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.44428694248199463,
+      "learning_rate": 9.612759575284483e-05,
+      "loss": 0.2933,
+      "step": 322
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4625035524368286,
+      "learning_rate": 9.609791636868281e-05,
+      "loss": 0.3519,
+      "step": 323
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.44164377450942993,
+      "learning_rate": 9.606812829949896e-05,
+      "loss": 0.3175,
+      "step": 324
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5235888957977295,
+      "learning_rate": 9.603823161552457e-05,
+      "loss": 0.4023,
+      "step": 325
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.46865010261535645,
+      "learning_rate": 9.600822638724705e-05,
+      "loss": 0.3122,
+      "step": 326
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4579242467880249,
+      "learning_rate": 9.597811268540968e-05,
+      "loss": 0.3616,
+      "step": 327
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4458639919757843,
+      "learning_rate": 9.594789058101153e-05,
+      "loss": 0.3751,
+      "step": 328
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4335557818412781,
+      "learning_rate": 9.591756014530722e-05,
+      "loss": 0.3695,
+      "step": 329
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.46497511863708496,
+      "learning_rate": 9.588712144980681e-05,
+      "loss": 0.3334,
+      "step": 330
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.4799109995365143,
+      "learning_rate": 9.585657456627556e-05,
+      "loss": 0.3296,
+      "step": 331
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.48978886008262634,
+      "learning_rate": 9.582591956673387e-05,
+      "loss": 0.4027,
+      "step": 332
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.529635488986969,
+      "learning_rate": 9.579515652345698e-05,
+      "loss": 0.3917,
+      "step": 333
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4566310942173004,
+      "learning_rate": 9.576428550897489e-05,
+      "loss": 0.2521,
+      "step": 334
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.45112448930740356,
+      "learning_rate": 9.57333065960722e-05,
+      "loss": 0.3677,
+      "step": 335
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4179379343986511,
+      "learning_rate": 9.570221985778785e-05,
+      "loss": 0.3077,
+      "step": 336
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.4628843665122986,
+      "learning_rate": 9.567102536741501e-05,
+      "loss": 0.3718,
+      "step": 337
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.46087244153022766,
+      "learning_rate": 9.563972319850092e-05,
+      "loss": 0.3679,
+      "step": 338
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.36096885800361633,
+      "learning_rate": 9.560831342484667e-05,
+      "loss": 0.2271,
+      "step": 339
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.48711246252059937,
+      "learning_rate": 9.557679612050708e-05,
+      "loss": 0.4142,
+      "step": 340
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3861476182937622,
+      "learning_rate": 9.554517135979043e-05,
+      "loss": 0.3286,
+      "step": 341
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.340451717376709,
+      "learning_rate": 9.551343921725843e-05,
+      "loss": 0.2001,
+      "step": 342
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5100023150444031,
+      "learning_rate": 9.548159976772592e-05,
+      "loss": 0.374,
+      "step": 343
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5192328095436096,
+      "learning_rate": 9.544965308626074e-05,
+      "loss": 0.394,
+      "step": 344
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3396227955818176,
+      "learning_rate": 9.541759924818357e-05,
+      "loss": 0.2617,
+      "step": 345
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5100813508033752,
+      "learning_rate": 9.538543832906773e-05,
+      "loss": 0.3553,
+      "step": 346
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4389209449291229,
+      "learning_rate": 9.535317040473895e-05,
+      "loss": 0.3837,
+      "step": 347
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.45119544863700867,
+      "learning_rate": 9.53207955512753e-05,
+      "loss": 0.3934,
+      "step": 348
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5094997882843018,
+      "learning_rate": 9.528831384500698e-05,
+      "loss": 0.3568,
+      "step": 349
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.39556166529655457,
+      "learning_rate": 9.525572536251607e-05,
+      "loss": 0.2934,
+      "step": 350
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.567779541015625,
+      "learning_rate": 9.52230301806364e-05,
+      "loss": 0.4005,
+      "step": 351
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.39256539940834045,
+      "learning_rate": 9.519022837645337e-05,
+      "loss": 0.2731,
+      "step": 352
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.4692497253417969,
+      "learning_rate": 9.51573200273038e-05,
+      "loss": 0.3482,
+      "step": 353
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4031953513622284,
+      "learning_rate": 9.512430521077565e-05,
+      "loss": 0.3191,
+      "step": 354
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4426116943359375,
+      "learning_rate": 9.509118400470793e-05,
+      "loss": 0.3683,
+      "step": 355
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.39099055528640747,
+      "learning_rate": 9.505795648719048e-05,
+      "loss": 0.3096,
+      "step": 356
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.4971387982368469,
+      "learning_rate": 9.50246227365638e-05,
+      "loss": 0.3297,
+      "step": 357
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.41271600127220154,
+      "learning_rate": 9.499118283141887e-05,
+      "loss": 0.3024,
+      "step": 358
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3604642152786255,
+      "learning_rate": 9.495763685059688e-05,
+      "loss": 0.2313,
+      "step": 359
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.35572880506515503,
+      "learning_rate": 9.492398487318921e-05,
+      "loss": 0.2456,
+      "step": 360
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.40534543991088867,
+      "learning_rate": 9.489022697853709e-05,
+      "loss": 0.2614,
+      "step": 361
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.425870805978775,
+      "learning_rate": 9.485636324623146e-05,
+      "loss": 0.3329,
+      "step": 362
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.45815330743789673,
+      "learning_rate": 9.482239375611282e-05,
+      "loss": 0.2862,
+      "step": 363
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.42662203311920166,
+      "learning_rate": 9.478831858827104e-05,
+      "loss": 0.285,
+      "step": 364
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4303944706916809,
+      "learning_rate": 9.475413782304507e-05,
+      "loss": 0.3598,
+      "step": 365
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.39889052510261536,
+      "learning_rate": 9.471985154102292e-05,
+      "loss": 0.2269,
+      "step": 366
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.31236928701400757,
+      "learning_rate": 9.468545982304132e-05,
+      "loss": 0.1666,
+      "step": 367
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4240448772907257,
+      "learning_rate": 9.465096275018556e-05,
+      "loss": 0.267,
+      "step": 368
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3831271827220917,
+      "learning_rate": 9.46163604037894e-05,
+      "loss": 0.2006,
+      "step": 369
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3652058243751526,
+      "learning_rate": 9.458165286543476e-05,
+      "loss": 0.2196,
+      "step": 370
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.365714967250824,
+      "learning_rate": 9.454684021695157e-05,
+      "loss": 0.2261,
+      "step": 371
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4304807484149933,
+      "learning_rate": 9.451192254041758e-05,
+      "loss": 0.262,
+      "step": 372
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.44848257303237915,
+      "learning_rate": 9.447689991815818e-05,
+      "loss": 0.2896,
+      "step": 373
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.40460604429244995,
+      "learning_rate": 9.444177243274618e-05,
+      "loss": 0.2193,
+      "step": 374
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5136522054672241,
+      "learning_rate": 9.440654016700161e-05,
+      "loss": 0.3631,
+      "step": 375
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5594468712806702,
+      "learning_rate": 9.437120320399158e-05,
+      "loss": 0.4571,
+      "step": 376
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5116495490074158,
+      "learning_rate": 9.433576162703e-05,
+      "loss": 0.3476,
+      "step": 377
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5234118700027466,
+      "learning_rate": 9.430021551967744e-05,
+      "loss": 0.3789,
+      "step": 378
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.45437514781951904,
+      "learning_rate": 9.426456496574096e-05,
+      "loss": 0.3603,
+      "step": 379
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4519050717353821,
+      "learning_rate": 9.422881004927381e-05,
+      "loss": 0.28,
+      "step": 380
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3754940330982208,
+      "learning_rate": 9.419295085457536e-05,
+      "loss": 0.282,
+      "step": 381
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.44995343685150146,
+      "learning_rate": 9.415698746619079e-05,
+      "loss": 0.3192,
+      "step": 382
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.40093666315078735,
+      "learning_rate": 9.412091996891096e-05,
+      "loss": 0.2968,
+      "step": 383
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.32370567321777344,
+      "learning_rate": 9.408474844777217e-05,
+      "loss": 0.2334,
+      "step": 384
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3858683407306671,
+      "learning_rate": 9.404847298805601e-05,
+      "loss": 0.3117,
+      "step": 385
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4255905747413635,
+      "learning_rate": 9.401209367528908e-05,
+      "loss": 0.3185,
+      "step": 386
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3737567067146301,
+      "learning_rate": 9.397561059524285e-05,
+      "loss": 0.2261,
+      "step": 387
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3858630657196045,
+      "learning_rate": 9.393902383393347e-05,
+      "loss": 0.2919,
+      "step": 388
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.4931787848472595,
+      "learning_rate": 9.39023334776215e-05,
+      "loss": 0.4583,
+      "step": 389
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.438981294631958,
+      "learning_rate": 9.386553961281179e-05,
+      "loss": 0.3528,
+      "step": 390
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.442809134721756,
+      "learning_rate": 9.382864232625322e-05,
+      "loss": 0.353,
+      "step": 391
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2885504364967346,
+      "learning_rate": 9.379164170493844e-05,
+      "loss": 0.2061,
+      "step": 392
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3465559482574463,
+      "learning_rate": 9.375453783610382e-05,
+      "loss": 0.2591,
+      "step": 393
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.39951983094215393,
+      "learning_rate": 9.371733080722911e-05,
+      "loss": 0.2874,
+      "step": 394
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3922140300273895,
+      "learning_rate": 9.36800207060373e-05,
+      "loss": 0.3395,
+      "step": 395
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3735324740409851,
+      "learning_rate": 9.36426076204944e-05,
+      "loss": 0.2285,
+      "step": 396
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.46303531527519226,
+      "learning_rate": 9.360509163880921e-05,
+      "loss": 0.3076,
+      "step": 397
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4106721878051758,
+      "learning_rate": 9.35674728494331e-05,
+      "loss": 0.2804,
+      "step": 398
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3216957747936249,
+      "learning_rate": 9.35297513410599e-05,
+      "loss": 0.2276,
+      "step": 399
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.48118698596954346,
+      "learning_rate": 9.349192720262555e-05,
+      "loss": 0.4019,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.3041338622570038,
+      "eval_runtime": 367.3376,
+      "eval_samples_per_second": 0.928,
+      "eval_steps_per_second": 0.234,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.38611844182014465,
+      "learning_rate": 9.345400052330802e-05,
+      "loss": 0.288,
+      "step": 401
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.37662482261657715,
+      "learning_rate": 9.341597139252698e-05,
+      "loss": 0.3073,
+      "step": 402
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.391392320394516,
+      "learning_rate": 9.337783989994371e-05,
+      "loss": 0.2835,
+      "step": 403
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.41128742694854736,
+      "learning_rate": 9.333960613546079e-05,
+      "loss": 0.3034,
+      "step": 404
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3846389353275299,
+      "learning_rate": 9.330127018922194e-05,
+      "loss": 0.3054,
+      "step": 405
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.37532809376716614,
+      "learning_rate": 9.326283215161178e-05,
+      "loss": 0.2446,
+      "step": 406
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4374251961708069,
+      "learning_rate": 9.322429211325566e-05,
+      "loss": 0.2701,
+      "step": 407
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.44123896956443787,
+      "learning_rate": 9.31856501650194e-05,
+      "loss": 0.3324,
+      "step": 408
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.39915063977241516,
+      "learning_rate": 9.314690639800906e-05,
+      "loss": 0.2815,
+      "step": 409
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.38047724962234497,
+      "learning_rate": 9.310806090357082e-05,
+      "loss": 0.2393,
+      "step": 410
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.35422012209892273,
+      "learning_rate": 9.306911377329066e-05,
+      "loss": 0.269,
+      "step": 411
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.492446631193161,
+      "learning_rate": 9.303006509899418e-05,
+      "loss": 0.348,
+      "step": 412
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3877454698085785,
+      "learning_rate": 9.299091497274647e-05,
+      "loss": 0.2814,
+      "step": 413
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3836115896701813,
+      "learning_rate": 9.29516634868517e-05,
+      "loss": 0.2714,
+      "step": 414
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.41554176807403564,
+      "learning_rate": 9.291231073385307e-05,
+      "loss": 0.3488,
+      "step": 415
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.4118254482746124,
+      "learning_rate": 9.287285680653254e-05,
+      "loss": 0.3286,
+      "step": 416
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3940303921699524,
+      "learning_rate": 9.283330179791063e-05,
+      "loss": 0.2859,
+      "step": 417
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4235418140888214,
+      "learning_rate": 9.279364580124614e-05,
+      "loss": 0.3051,
+      "step": 418
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.45912712812423706,
+      "learning_rate": 9.275388891003596e-05,
+      "loss": 0.3476,
+      "step": 419
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3564731180667877,
+      "learning_rate": 9.27140312180149e-05,
+      "loss": 0.2353,
+      "step": 420
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.43604397773742676,
+      "learning_rate": 9.267407281915541e-05,
+      "loss": 0.3049,
+      "step": 421
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4366002678871155,
+      "learning_rate": 9.263401380766739e-05,
+      "loss": 0.347,
+      "step": 422
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.39401495456695557,
+      "learning_rate": 9.25938542779979e-05,
+      "loss": 0.2877,
+      "step": 423
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.35046637058258057,
+      "learning_rate": 9.255359432483105e-05,
+      "loss": 0.2383,
+      "step": 424
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.43969568610191345,
+      "learning_rate": 9.251323404308774e-05,
+      "loss": 0.3215,
+      "step": 425
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.41940975189208984,
+      "learning_rate": 9.247277352792533e-05,
+      "loss": 0.3408,
+      "step": 426
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.36807939410209656,
+      "learning_rate": 9.243221287473756e-05,
+      "loss": 0.2945,
+      "step": 427
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4048173725605011,
+      "learning_rate": 9.239155217915422e-05,
+      "loss": 0.3122,
+      "step": 428
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.42843782901763916,
+      "learning_rate": 9.235079153704107e-05,
+      "loss": 0.3497,
+      "step": 429
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3625134229660034,
+      "learning_rate": 9.230993104449939e-05,
+      "loss": 0.2124,
+      "step": 430
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3488845229148865,
+      "learning_rate": 9.226897079786593e-05,
+      "loss": 0.2953,
+      "step": 431
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4257933497428894,
+      "learning_rate": 9.222791089371265e-05,
+      "loss": 0.3617,
+      "step": 432
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4866938591003418,
+      "learning_rate": 9.218675142884647e-05,
+      "loss": 0.3947,
+      "step": 433
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.474409282207489,
+      "learning_rate": 9.214549250030899e-05,
+      "loss": 0.443,
+      "step": 434
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.39310964941978455,
+      "learning_rate": 9.210413420537638e-05,
+      "loss": 0.3048,
+      "step": 435
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5172072649002075,
+      "learning_rate": 9.206267664155907e-05,
+      "loss": 0.3862,
+      "step": 436
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4572441577911377,
+      "learning_rate": 9.20211199066015e-05,
+      "loss": 0.3488,
+      "step": 437
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4569890797138214,
+      "learning_rate": 9.197946409848194e-05,
+      "loss": 0.4029,
+      "step": 438
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.40921786427497864,
+      "learning_rate": 9.19377093154123e-05,
+      "loss": 0.3472,
+      "step": 439
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4346557855606079,
+      "learning_rate": 9.189585565583778e-05,
+      "loss": 0.3326,
+      "step": 440
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.41340482234954834,
+      "learning_rate": 9.185390321843672e-05,
+      "loss": 0.3002,
+      "step": 441
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.33492302894592285,
+      "learning_rate": 9.181185210212034e-05,
+      "loss": 0.1826,
+      "step": 442
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.510562539100647,
+      "learning_rate": 9.176970240603253e-05,
+      "loss": 0.4325,
+      "step": 443
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.39474889636039734,
+      "learning_rate": 9.172745422954961e-05,
+      "loss": 0.2834,
+      "step": 444
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3410179316997528,
+      "learning_rate": 9.168510767228007e-05,
+      "loss": 0.2178,
+      "step": 445
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.42597004771232605,
+      "learning_rate": 9.164266283406433e-05,
+      "loss": 0.2536,
+      "step": 446
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4418257474899292,
+      "learning_rate": 9.160011981497457e-05,
+      "loss": 0.37,
+      "step": 447
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.4709400534629822,
+      "learning_rate": 9.155747871531444e-05,
+      "loss": 0.3232,
+      "step": 448
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.48430338501930237,
+      "learning_rate": 9.151473963561883e-05,
+      "loss": 0.3848,
+      "step": 449
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4547795057296753,
+      "learning_rate": 9.14719026766536e-05,
+      "loss": 0.3819,
+      "step": 450
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7274636626243591,
+      "learning_rate": 9.142896793941546e-05,
+      "loss": 0.4455,
+      "step": 451
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4149537682533264,
+      "learning_rate": 9.13859355251316e-05,
+      "loss": 0.3043,
+      "step": 452
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3654284179210663,
+      "learning_rate": 9.134280553525946e-05,
+      "loss": 0.2751,
+      "step": 453
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.45339077711105347,
+      "learning_rate": 9.129957807148666e-05,
+      "loss": 0.3266,
+      "step": 454
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.45142748951911926,
+      "learning_rate": 9.12562532357305e-05,
+      "loss": 0.3467,
+      "step": 455
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3450026512145996,
+      "learning_rate": 9.121283113013795e-05,
+      "loss": 0.2091,
+      "step": 456
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.4695757329463959,
+      "learning_rate": 9.116931185708523e-05,
+      "loss": 0.352,
+      "step": 457
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3936323821544647,
+      "learning_rate": 9.112569551917774e-05,
+      "loss": 0.2727,
+      "step": 458
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.421451210975647,
+      "learning_rate": 9.108198221924966e-05,
+      "loss": 0.3032,
+      "step": 459
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.36577218770980835,
+      "learning_rate": 9.103817206036382e-05,
+      "loss": 0.2387,
+      "step": 460
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.42635923624038696,
+      "learning_rate": 9.09942651458114e-05,
+      "loss": 0.3572,
+      "step": 461
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4160636067390442,
+      "learning_rate": 9.095026157911166e-05,
+      "loss": 0.2632,
+      "step": 462
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4471433758735657,
+      "learning_rate": 9.090616146401183e-05,
+      "loss": 0.4012,
+      "step": 463
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.41552168130874634,
+      "learning_rate": 9.086196490448668e-05,
+      "loss": 0.277,
+      "step": 464
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.36891818046569824,
+      "learning_rate": 9.081767200473842e-05,
+      "loss": 0.253,
+      "step": 465
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4040123224258423,
+      "learning_rate": 9.077328286919638e-05,
+      "loss": 0.2875,
+      "step": 466
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4402499794960022,
+      "learning_rate": 9.072879760251679e-05,
+      "loss": 0.328,
+      "step": 467
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.4268719255924225,
+      "learning_rate": 9.068421630958254e-05,
+      "loss": 0.3059,
+      "step": 468
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.37686699628829956,
+      "learning_rate": 9.06395390955029e-05,
+      "loss": 0.2735,
+      "step": 469
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4133022725582123,
+      "learning_rate": 9.059476606561328e-05,
+      "loss": 0.2951,
+      "step": 470
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3832361400127411,
+      "learning_rate": 9.054989732547506e-05,
+      "loss": 0.3092,
+      "step": 471
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.415218710899353,
+      "learning_rate": 9.050493298087523e-05,
+      "loss": 0.3727,
+      "step": 472
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.526370108127594,
+      "learning_rate": 9.045987313782616e-05,
+      "loss": 0.4884,
+      "step": 473
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.37246769666671753,
+      "learning_rate": 9.041471790256542e-05,
+      "loss": 0.2333,
+      "step": 474
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3858553469181061,
+      "learning_rate": 9.036946738155548e-05,
+      "loss": 0.2428,
+      "step": 475
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4015858471393585,
+      "learning_rate": 9.032412168148345e-05,
+      "loss": 0.3068,
+      "step": 476
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4636778235435486,
+      "learning_rate": 9.027868090926088e-05,
+      "loss": 0.3447,
+      "step": 477
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3941800892353058,
+      "learning_rate": 9.02331451720234e-05,
+      "loss": 0.3276,
+      "step": 478
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3637751042842865,
+      "learning_rate": 9.018751457713062e-05,
+      "loss": 0.2647,
+      "step": 479
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4784156084060669,
+      "learning_rate": 9.014178923216572e-05,
+      "loss": 0.3795,
+      "step": 480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3805742561817169,
+      "learning_rate": 9.009596924493535e-05,
+      "loss": 0.3076,
+      "step": 481
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.39622727036476135,
+      "learning_rate": 9.005005472346924e-05,
+      "loss": 0.2892,
+      "step": 482
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5285367369651794,
+      "learning_rate": 9.000404577602002e-05,
+      "loss": 0.4502,
+      "step": 483
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.4529123902320862,
+      "learning_rate": 8.995794251106295e-05,
+      "loss": 0.3764,
+      "step": 484
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.398775190114975,
+      "learning_rate": 8.991174503729567e-05,
+      "loss": 0.3045,
+      "step": 485
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.39205294847488403,
+      "learning_rate": 8.986545346363792e-05,
+      "loss": 0.2996,
+      "step": 486
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.488794207572937,
+      "learning_rate": 8.98190678992313e-05,
+      "loss": 0.3167,
+      "step": 487
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5263704657554626,
+      "learning_rate": 8.977258845343905e-05,
+      "loss": 0.4342,
+      "step": 488
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.45423367619514465,
+      "learning_rate": 8.97260152358457e-05,
+      "loss": 0.3566,
+      "step": 489
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.34878504276275635,
+      "learning_rate": 8.967934835625689e-05,
+      "loss": 0.2469,
+      "step": 490
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4419337511062622,
+      "learning_rate": 8.963258792469908e-05,
+      "loss": 0.3786,
+      "step": 491
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.42417728900909424,
+      "learning_rate": 8.958573405141932e-05,
+      "loss": 0.3307,
+      "step": 492
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.44068655371665955,
+      "learning_rate": 8.953878684688493e-05,
+      "loss": 0.3085,
+      "step": 493
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4266207814216614,
+      "learning_rate": 8.949174642178332e-05,
+      "loss": 0.3054,
+      "step": 494
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.35465604066848755,
+      "learning_rate": 8.944461288702166e-05,
+      "loss": 0.2471,
+      "step": 495
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.37964946031570435,
+      "learning_rate": 8.939738635372664e-05,
+      "loss": 0.2909,
+      "step": 496
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4431998133659363,
+      "learning_rate": 8.935006693324423e-05,
+      "loss": 0.3073,
+      "step": 497
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.4371538460254669,
+      "learning_rate": 8.930265473713938e-05,
+      "loss": 0.3963,
+      "step": 498
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.43692684173583984,
+      "learning_rate": 8.925514987719579e-05,
+      "loss": 0.3383,
+      "step": 499
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3857593238353729,
+      "learning_rate": 8.920755246541563e-05,
+      "loss": 0.2701,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.3033298850059509,
+      "eval_runtime": 366.9528,
+      "eval_samples_per_second": 0.929,
+      "eval_steps_per_second": 0.234,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5027936100959778,
+      "learning_rate": 8.915986261401931e-05,
+      "loss": 0.4645,
+      "step": 501
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.34565359354019165,
+      "learning_rate": 8.911208043544513e-05,
+      "loss": 0.2196,
+      "step": 502
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.46481138467788696,
+      "learning_rate": 8.906420604234908e-05,
+      "loss": 0.3876,
+      "step": 503
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4338441789150238,
+      "learning_rate": 8.90162395476046e-05,
+      "loss": 0.3684,
+      "step": 504
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4877359867095947,
+      "learning_rate": 8.896818106430224e-05,
+      "loss": 0.3709,
+      "step": 505
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4262893795967102,
+      "learning_rate": 8.89200307057495e-05,
+      "loss": 0.2752,
+      "step": 506
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3778301775455475,
+      "learning_rate": 8.887178858547039e-05,
+      "loss": 0.2889,
+      "step": 507
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4494069218635559,
+      "learning_rate": 8.882345481720533e-05,
+      "loss": 0.2943,
+      "step": 508
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.41227301955223083,
+      "learning_rate": 8.877502951491082e-05,
+      "loss": 0.2774,
+      "step": 509
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.37210455536842346,
+      "learning_rate": 8.872651279275916e-05,
+      "loss": 0.2971,
+      "step": 510
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.4338979721069336,
+      "learning_rate": 8.867790476513818e-05,
+      "loss": 0.2316,
+      "step": 511
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.419693261384964,
+      "learning_rate": 8.862920554665098e-05,
+      "loss": 0.2713,
+      "step": 512
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.42196741700172424,
+      "learning_rate": 8.858041525211569e-05,
+      "loss": 0.293,
+      "step": 513
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5517318844795227,
+      "learning_rate": 8.853153399656513e-05,
+      "loss": 0.3999,
+      "step": 514
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.44009464979171753,
+      "learning_rate": 8.848256189524662e-05,
+      "loss": 0.3185,
+      "step": 515
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4338436424732208,
+      "learning_rate": 8.843349906362163e-05,
+      "loss": 0.3291,
+      "step": 516
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5298638939857483,
+      "learning_rate": 8.838434561736556e-05,
+      "loss": 0.3703,
+      "step": 517
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4447985291481018,
+      "learning_rate": 8.833510167236747e-05,
+      "loss": 0.2848,
+      "step": 518
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3730708658695221,
+      "learning_rate": 8.828576734472974e-05,
+      "loss": 0.252,
+      "step": 519
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5157079100608826,
+      "learning_rate": 8.823634275076791e-05,
+      "loss": 0.4092,
+      "step": 520
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.44079431891441345,
+      "learning_rate": 8.818682800701029e-05,
+      "loss": 0.2978,
+      "step": 521
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.44436514377593994,
+      "learning_rate": 8.813722323019773e-05,
+      "loss": 0.3527,
+      "step": 522
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4249900281429291,
+      "learning_rate": 8.808752853728341e-05,
+      "loss": 0.2741,
+      "step": 523
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3489685356616974,
+      "learning_rate": 8.803774404543247e-05,
+      "loss": 0.2173,
+      "step": 524
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.46010851860046387,
+      "learning_rate": 8.798786987202174e-05,
+      "loss": 0.3313,
+      "step": 525
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.47715631127357483,
+      "learning_rate": 8.793790613463955e-05,
+      "loss": 0.334,
+      "step": 526
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.46374446153640747,
+      "learning_rate": 8.788785295108535e-05,
+      "loss": 0.3277,
+      "step": 527
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.47940391302108765,
+      "learning_rate": 8.783771043936948e-05,
+      "loss": 0.3523,
+      "step": 528
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4219704866409302,
+      "learning_rate": 8.778747871771292e-05,
+      "loss": 0.2992,
+      "step": 529
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4081328809261322,
+      "learning_rate": 8.773715790454695e-05,
+      "loss": 0.2879,
+      "step": 530
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.41193005442619324,
+      "learning_rate": 8.768674811851293e-05,
+      "loss": 0.3539,
+      "step": 531
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4311436712741852,
+      "learning_rate": 8.763624947846195e-05,
+      "loss": 0.3307,
+      "step": 532
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.37866875529289246,
+      "learning_rate": 8.758566210345462e-05,
+      "loss": 0.2866,
+      "step": 533
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.42311981320381165,
+      "learning_rate": 8.753498611276079e-05,
+      "loss": 0.2248,
+      "step": 534
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.46103572845458984,
+      "learning_rate": 8.748422162585914e-05,
+      "loss": 0.4649,
+      "step": 535
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.39613980054855347,
+      "learning_rate": 8.743336876243712e-05,
+      "loss": 0.2812,
+      "step": 536
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.40790408849716187,
+      "learning_rate": 8.738242764239046e-05,
+      "loss": 0.3281,
+      "step": 537
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4292603135108948,
+      "learning_rate": 8.7331398385823e-05,
+      "loss": 0.3481,
+      "step": 538
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4079511761665344,
+      "learning_rate": 8.728028111304638e-05,
+      "loss": 0.2985,
+      "step": 539
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4475175142288208,
+      "learning_rate": 8.722907594457974e-05,
+      "loss": 0.3546,
+      "step": 540
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4934547543525696,
+      "learning_rate": 8.717778300114951e-05,
+      "loss": 0.3598,
+      "step": 541
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.470649391412735,
+      "learning_rate": 8.712640240368899e-05,
+      "loss": 0.3453,
+      "step": 542
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.38042715191841125,
+      "learning_rate": 8.707493427333818e-05,
+      "loss": 0.2648,
+      "step": 543
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.42057284712791443,
+      "learning_rate": 8.702337873144343e-05,
+      "loss": 0.3377,
+      "step": 544
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.40261784195899963,
+      "learning_rate": 8.697173589955724e-05,
+      "loss": 0.3497,
+      "step": 545
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3466137945652008,
+      "learning_rate": 8.692000589943785e-05,
+      "loss": 0.2408,
+      "step": 546
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4354414641857147,
+      "learning_rate": 8.686818885304905e-05,
+      "loss": 0.3668,
+      "step": 547
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.44051098823547363,
+      "learning_rate": 8.681628488255986e-05,
+      "loss": 0.2501,
+      "step": 548
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.44984376430511475,
+      "learning_rate": 8.676429411034422e-05,
+      "loss": 0.366,
+      "step": 549
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4447495639324188,
+      "learning_rate": 8.671221665898073e-05,
+      "loss": 0.2846,
+      "step": 550
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3866206109523773,
+      "learning_rate": 8.666005265125236e-05,
+      "loss": 0.2442,
+      "step": 551
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4412643611431122,
+      "learning_rate": 8.660780221014616e-05,
+      "loss": 0.3094,
+      "step": 552
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.48346805572509766,
+      "learning_rate": 8.655546545885294e-05,
+      "loss": 0.4002,
+      "step": 553
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4209979772567749,
+      "learning_rate": 8.650304252076703e-05,
+      "loss": 0.2935,
+      "step": 554
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4458561837673187,
+      "learning_rate": 8.645053351948593e-05,
+      "loss": 0.3018,
+      "step": 555
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3844420909881592,
+      "learning_rate": 8.639793857881011e-05,
+      "loss": 0.2815,
+      "step": 556
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.49796560406684875,
+      "learning_rate": 8.634525782274258e-05,
+      "loss": 0.4174,
+      "step": 557
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.528649628162384,
+      "learning_rate": 8.629249137548874e-05,
+      "loss": 0.3942,
+      "step": 558
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4881141781806946,
+      "learning_rate": 8.6239639361456e-05,
+      "loss": 0.3795,
+      "step": 559
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5525495409965515,
+      "learning_rate": 8.618670190525352e-05,
+      "loss": 0.4617,
+      "step": 560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.37910011410713196,
+      "learning_rate": 8.613367913169187e-05,
+      "loss": 0.3065,
+      "step": 561
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.38476353883743286,
+      "learning_rate": 8.608057116578283e-05,
+      "loss": 0.1864,
+      "step": 562
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.4849575459957123,
+      "learning_rate": 8.602737813273901e-05,
+      "loss": 0.3469,
+      "step": 563
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3707044720649719,
+      "learning_rate": 8.597410015797358e-05,
+      "loss": 0.2221,
+      "step": 564
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4276258051395416,
+      "learning_rate": 8.592073736709996e-05,
+      "loss": 0.274,
+      "step": 565
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4069690704345703,
+      "learning_rate": 8.586728988593158e-05,
+      "loss": 0.2044,
+      "step": 566
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4143349230289459,
+      "learning_rate": 8.581375784048153e-05,
+      "loss": 0.2899,
+      "step": 567
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.42278823256492615,
+      "learning_rate": 8.576014135696226e-05,
+      "loss": 0.2703,
+      "step": 568
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.49254170060157776,
+      "learning_rate": 8.570644056178532e-05,
+      "loss": 0.3761,
+      "step": 569
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.48147767782211304,
+      "learning_rate": 8.565265558156101e-05,
+      "loss": 0.3444,
+      "step": 570
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4823344945907593,
+      "learning_rate": 8.559878654309819e-05,
+      "loss": 0.3627,
+      "step": 571
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.34366798400878906,
+      "learning_rate": 8.554483357340379e-05,
+      "loss": 0.2068,
+      "step": 572
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5252037644386292,
+      "learning_rate": 8.549079679968272e-05,
+      "loss": 0.4436,
+      "step": 573
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.44619235396385193,
+      "learning_rate": 8.543667634933742e-05,
+      "loss": 0.3824,
+      "step": 574
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3972527086734772,
+      "learning_rate": 8.538247234996766e-05,
+      "loss": 0.2556,
+      "step": 575
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5212999582290649,
+      "learning_rate": 8.532818492937014e-05,
+      "loss": 0.4071,
+      "step": 576
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.28184422850608826,
+      "learning_rate": 8.527381421553829e-05,
+      "loss": 0.1877,
+      "step": 577
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5419026613235474,
+      "learning_rate": 8.521936033666187e-05,
+      "loss": 0.2142,
+      "step": 578
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4146403670310974,
+      "learning_rate": 8.516482342112679e-05,
+      "loss": 0.3272,
+      "step": 579
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3827855885028839,
+      "learning_rate": 8.511020359751466e-05,
+      "loss": 0.2447,
+      "step": 580
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3556893467903137,
+      "learning_rate": 8.505550099460265e-05,
+      "loss": 0.2609,
+      "step": 581
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4643717110157013,
+      "learning_rate": 8.500071574136295e-05,
+      "loss": 0.3811,
+      "step": 582
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.40621811151504517,
+      "learning_rate": 8.49458479669628e-05,
+      "loss": 0.3068,
+      "step": 583
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.43951380252838135,
+      "learning_rate": 8.489089780076386e-05,
+      "loss": 0.2911,
+      "step": 584
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4206659197807312,
+      "learning_rate": 8.483586537232211e-05,
+      "loss": 0.3139,
+      "step": 585
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4267883598804474,
+      "learning_rate": 8.478075081138745e-05,
+      "loss": 0.3283,
+      "step": 586
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3936096131801605,
+      "learning_rate": 8.472555424790347e-05,
+      "loss": 0.287,
+      "step": 587
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.47112205624580383,
+      "learning_rate": 8.467027581200702e-05,
+      "loss": 0.3587,
+      "step": 588
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.36312347650527954,
+      "learning_rate": 8.461491563402807e-05,
+      "loss": 0.2283,
+      "step": 589
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.38688814640045166,
+      "learning_rate": 8.455947384448924e-05,
+      "loss": 0.2529,
+      "step": 590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4152275323867798,
+      "learning_rate": 8.450395057410561e-05,
+      "loss": 0.2906,
+      "step": 591
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.35978758335113525,
+      "learning_rate": 8.444834595378434e-05,
+      "loss": 0.2366,
+      "step": 592
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.47361108660697937,
+      "learning_rate": 8.43926601146244e-05,
+      "loss": 0.2986,
+      "step": 593
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.41231879591941833,
+      "learning_rate": 8.433689318791626e-05,
+      "loss": 0.3025,
+      "step": 594
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.4158206582069397,
+      "learning_rate": 8.428104530514154e-05,
+      "loss": 0.2844,
+      "step": 595
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.42700937390327454,
+      "learning_rate": 8.422511659797279e-05,
+      "loss": 0.2894,
+      "step": 596
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.44759729504585266,
+      "learning_rate": 8.416910719827304e-05,
+      "loss": 0.3687,
+      "step": 597
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3688613474369049,
+      "learning_rate": 8.411301723809563e-05,
+      "loss": 0.2606,
+      "step": 598
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4526158571243286,
+      "learning_rate": 8.405684684968383e-05,
+      "loss": 0.3255,
+      "step": 599
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4307437837123871,
+      "learning_rate": 8.400059616547047e-05,
+      "loss": 0.3165,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.30199378728866577,
+      "eval_runtime": 368.2619,
+      "eval_samples_per_second": 0.926,
+      "eval_steps_per_second": 0.234,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5042355060577393,
+      "learning_rate": 8.394426531807778e-05,
+      "loss": 0.4452,
+      "step": 601
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.35113629698753357,
+      "learning_rate": 8.388785444031695e-05,
+      "loss": 0.247,
+      "step": 602
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.42139703035354614,
+      "learning_rate": 8.383136366518788e-05,
+      "loss": 0.2978,
+      "step": 603
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4393561780452728,
+      "learning_rate": 8.377479312587879e-05,
+      "loss": 0.2966,
+      "step": 604
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5458940267562866,
+      "learning_rate": 8.371814295576603e-05,
+      "loss": 0.4765,
+      "step": 605
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 8.880595207214355,
+      "learning_rate": 8.366141328841366e-05,
+      "loss": 0.5082,
+      "step": 606
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5241791605949402,
+      "learning_rate": 8.360460425757314e-05,
+      "loss": 0.4453,
+      "step": 607
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4780680239200592,
+      "learning_rate": 8.354771599718312e-05,
+      "loss": 0.3079,
+      "step": 608
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39361393451690674,
+      "learning_rate": 8.349074864136898e-05,
+      "loss": 0.3217,
+      "step": 609
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39532896876335144,
+      "learning_rate": 8.343370232444261e-05,
+      "loss": 0.2797,
+      "step": 610
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3550987243652344,
+      "learning_rate": 8.33765771809021e-05,
+      "loss": 0.2572,
+      "step": 611
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38473519682884216,
+      "learning_rate": 8.331937334543132e-05,
+      "loss": 0.2502,
+      "step": 612
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.38292405009269714,
+      "learning_rate": 8.326209095289972e-05,
+      "loss": 0.286,
+      "step": 613
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4829404652118683,
+      "learning_rate": 8.320473013836196e-05,
+      "loss": 0.3823,
+      "step": 614
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3962880074977875,
+      "learning_rate": 8.314729103705758e-05,
+      "loss": 0.2875,
+      "step": 615
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.31709444522857666,
+      "learning_rate": 8.308977378441072e-05,
+      "loss": 0.2102,
+      "step": 616
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.39247071743011475,
+      "learning_rate": 8.303217851602973e-05,
+      "loss": 0.3118,
+      "step": 617
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3811770975589752,
+      "learning_rate": 8.297450536770697e-05,
+      "loss": 0.2772,
+      "step": 618
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3957882225513458,
+      "learning_rate": 8.291675447541833e-05,
+      "loss": 0.2739,
+      "step": 619
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3813614249229431,
+      "learning_rate": 8.28589259753231e-05,
+      "loss": 0.2448,
+      "step": 620
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.48293402791023254,
+      "learning_rate": 8.280102000376345e-05,
+      "loss": 0.3601,
+      "step": 621
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4571930170059204,
+      "learning_rate": 8.274303669726426e-05,
+      "loss": 0.438,
+      "step": 622
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4715982973575592,
+      "learning_rate": 8.268497619253273e-05,
+      "loss": 0.3693,
+      "step": 623
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3570612370967865,
+      "learning_rate": 8.262683862645804e-05,
+      "loss": 0.2181,
+      "step": 624
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.45369642972946167,
+      "learning_rate": 8.256862413611113e-05,
+      "loss": 0.4171,
+      "step": 625
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.39541661739349365,
+      "learning_rate": 8.251033285874421e-05,
+      "loss": 0.2571,
+      "step": 626
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.43222755193710327,
+      "learning_rate": 8.245196493179061e-05,
+      "loss": 0.381,
+      "step": 627
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.41552790999412537,
+      "learning_rate": 8.239352049286435e-05,
+      "loss": 0.3327,
+      "step": 628
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.38137102127075195,
+      "learning_rate": 8.233499967975982e-05,
+      "loss": 0.2807,
+      "step": 629
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.40857887268066406,
+      "learning_rate": 8.22764026304515e-05,
+      "loss": 0.2715,
+      "step": 630
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4507109820842743,
+      "learning_rate": 8.221772948309362e-05,
+      "loss": 0.3949,
+      "step": 631
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4330926835536957,
+      "learning_rate": 8.21589803760198e-05,
+      "loss": 0.3471,
+      "step": 632
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.41828611493110657,
+      "learning_rate": 8.210015544774278e-05,
+      "loss": 0.3741,
+      "step": 633
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3408172130584717,
+      "learning_rate": 8.204125483695403e-05,
+      "loss": 0.2244,
+      "step": 634
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.48684483766555786,
+      "learning_rate": 8.198227868252348e-05,
+      "loss": 0.3867,
+      "step": 635
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.431608647108078,
+      "learning_rate": 8.192322712349917e-05,
+      "loss": 0.3075,
+      "step": 636
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4474915564060211,
+      "learning_rate": 8.186410029910694e-05,
+      "loss": 0.3787,
+      "step": 637
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4784719944000244,
+      "learning_rate": 8.180489834875e-05,
+      "loss": 0.3658,
+      "step": 638
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4535321593284607,
+      "learning_rate": 8.174562141200878e-05,
+      "loss": 0.3229,
+      "step": 639
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.39979875087738037,
+      "learning_rate": 8.168626962864045e-05,
+      "loss": 0.2434,
+      "step": 640
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.431804895401001,
+      "learning_rate": 8.16268431385787e-05,
+      "loss": 0.36,
+      "step": 641
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.45677754282951355,
+      "learning_rate": 8.156734208193327e-05,
+      "loss": 0.338,
+      "step": 642
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4533159136772156,
+      "learning_rate": 8.15077665989898e-05,
+      "loss": 0.3653,
+      "step": 643
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3977563977241516,
+      "learning_rate": 8.144811683020932e-05,
+      "loss": 0.3208,
+      "step": 644
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.45302459597587585,
+      "learning_rate": 8.138839291622808e-05,
+      "loss": 0.3477,
+      "step": 645
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.46249890327453613,
+      "learning_rate": 8.132859499785707e-05,
+      "loss": 0.3354,
+      "step": 646
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.42682644724845886,
+      "learning_rate": 8.126872321608184e-05,
+      "loss": 0.3098,
+      "step": 647
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.4579174816608429,
+      "learning_rate": 8.120877771206201e-05,
+      "loss": 0.3537,
+      "step": 648
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.46789848804473877,
+      "learning_rate": 8.114875862713106e-05,
+      "loss": 0.4004,
+      "step": 649
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.43281278014183044,
+      "learning_rate": 8.108866610279593e-05,
+      "loss": 0.3053,
+      "step": 650
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.49906638264656067,
+      "learning_rate": 8.102850028073674e-05,
+      "loss": 0.3026,
+      "step": 651
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3694945275783539,
+      "learning_rate": 8.096826130280639e-05,
+      "loss": 0.2052,
+      "step": 652
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.46960631012916565,
+      "learning_rate": 8.090794931103026e-05,
+      "loss": 0.2853,
+      "step": 653
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4513627886772156,
+      "learning_rate": 8.08475644476059e-05,
+      "loss": 0.3629,
+      "step": 654
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4009605646133423,
+      "learning_rate": 8.078710685490265e-05,
+      "loss": 0.2109,
+      "step": 655
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.42705237865448,
+      "learning_rate": 8.072657667546136e-05,
+      "loss": 0.3187,
+      "step": 656
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5091983079910278,
+      "learning_rate": 8.066597405199392e-05,
+      "loss": 0.3248,
+      "step": 657
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.4275987446308136,
+      "learning_rate": 8.060529912738315e-05,
+      "loss": 0.3091,
+      "step": 658
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.36369234323501587,
+      "learning_rate": 8.054455204468224e-05,
+      "loss": 0.202,
+      "step": 659
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.41810745000839233,
+      "learning_rate": 8.048373294711455e-05,
+      "loss": 0.2627,
+      "step": 660
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4977913498878479,
+      "learning_rate": 8.042284197807324e-05,
+      "loss": 0.3165,
+      "step": 661
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4297676980495453,
+      "learning_rate": 8.036187928112087e-05,
+      "loss": 0.2844,
+      "step": 662
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.44592076539993286,
+      "learning_rate": 8.030084499998916e-05,
+      "loss": 0.2857,
+      "step": 663
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5014137625694275,
+      "learning_rate": 8.023973927857857e-05,
+      "loss": 0.381,
+      "step": 664
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4190405309200287,
+      "learning_rate": 8.017856226095804e-05,
+      "loss": 0.2869,
+      "step": 665
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3442036211490631,
+      "learning_rate": 8.011731409136454e-05,
+      "loss": 0.1992,
+      "step": 666
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4580501914024353,
+      "learning_rate": 8.005599491420288e-05,
+      "loss": 0.3333,
+      "step": 667
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4559895694255829,
+      "learning_rate": 7.99946048740452e-05,
+      "loss": 0.3283,
+      "step": 668
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.419212281703949,
+      "learning_rate": 7.993314411563075e-05,
+      "loss": 0.2659,
+      "step": 669
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.41456085443496704,
+      "learning_rate": 7.987161278386554e-05,
+      "loss": 0.2893,
+      "step": 670
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4753405451774597,
+      "learning_rate": 7.981001102382192e-05,
+      "loss": 0.4192,
+      "step": 671
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3795643150806427,
+      "learning_rate": 7.974833898073832e-05,
+      "loss": 0.2777,
+      "step": 672
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5017170906066895,
+      "learning_rate": 7.968659680001886e-05,
+      "loss": 0.3473,
+      "step": 673
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.40297964215278625,
+      "learning_rate": 7.962478462723307e-05,
+      "loss": 0.2635,
+      "step": 674
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4292650818824768,
+      "learning_rate": 7.95629026081154e-05,
+      "loss": 0.35,
+      "step": 675
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3624914586544037,
+      "learning_rate": 7.950095088856508e-05,
+      "loss": 0.2652,
+      "step": 676
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4304388463497162,
+      "learning_rate": 7.943892961464563e-05,
+      "loss": 0.3278,
+      "step": 677
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4542439579963684,
+      "learning_rate": 7.937683893258454e-05,
+      "loss": 0.3535,
+      "step": 678
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3505571484565735,
+      "learning_rate": 7.931467898877297e-05,
+      "loss": 0.2276,
+      "step": 679
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4299335181713104,
+      "learning_rate": 7.925244992976538e-05,
+      "loss": 0.2793,
+      "step": 680
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4120853543281555,
+      "learning_rate": 7.919015190227919e-05,
+      "loss": 0.2984,
+      "step": 681
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.37679335474967957,
+      "learning_rate": 7.912778505319436e-05,
+      "loss": 0.2481,
+      "step": 682
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.45651429891586304,
+      "learning_rate": 7.906534952955321e-05,
+      "loss": 0.3639,
+      "step": 683
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.40571317076683044,
+      "learning_rate": 7.900284547855991e-05,
+      "loss": 0.2377,
+      "step": 684
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.43117037415504456,
+      "learning_rate": 7.894027304758022e-05,
+      "loss": 0.3017,
+      "step": 685
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5254279375076294,
+      "learning_rate": 7.88776323841411e-05,
+      "loss": 0.4287,
+      "step": 686
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.47938308119773865,
+      "learning_rate": 7.881492363593039e-05,
+      "loss": 0.352,
+      "step": 687
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4461328983306885,
+      "learning_rate": 7.875214695079647e-05,
+      "loss": 0.3569,
+      "step": 688
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5136838555335999,
+      "learning_rate": 7.868930247674788e-05,
+      "loss": 0.3784,
+      "step": 689
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4655466377735138,
+      "learning_rate": 7.862639036195298e-05,
+      "loss": 0.3199,
+      "step": 690
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.4025501012802124,
+      "learning_rate": 7.856341075473962e-05,
+      "loss": 0.2661,
+      "step": 691
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4783627390861511,
+      "learning_rate": 7.850036380359478e-05,
+      "loss": 0.3705,
+      "step": 692
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.484117329120636,
+      "learning_rate": 7.843724965716419e-05,
+      "loss": 0.3072,
+      "step": 693
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3950745463371277,
+      "learning_rate": 7.837406846425204e-05,
+      "loss": 0.2724,
+      "step": 694
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.45935970544815063,
+      "learning_rate": 7.831082037382057e-05,
+      "loss": 0.3437,
+      "step": 695
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.487434446811676,
+      "learning_rate": 7.824750553498976e-05,
+      "loss": 0.395,
+      "step": 696
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3948221802711487,
+      "learning_rate": 7.818412409703695e-05,
+      "loss": 0.2322,
+      "step": 697
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4208463728427887,
+      "learning_rate": 7.812067620939652e-05,
+      "loss": 0.2811,
+      "step": 698
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.4702443480491638,
+      "learning_rate": 7.805716202165949e-05,
+      "loss": 0.2904,
+      "step": 699
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5112441182136536,
+      "learning_rate": 7.799358168357323e-05,
+      "loss": 0.3771,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.29883939027786255,
+      "eval_runtime": 374.0297,
+      "eval_samples_per_second": 0.912,
+      "eval_steps_per_second": 0.23,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.44988197088241577,
+      "learning_rate": 7.792993534504102e-05,
+      "loss": 0.353,
+      "step": 701
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.38583260774612427,
+      "learning_rate": 7.786622315612183e-05,
+      "loss": 0.2196,
+      "step": 702
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4170338809490204,
+      "learning_rate": 7.78024452670298e-05,
+      "loss": 0.1996,
+      "step": 703
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3776649236679077,
+      "learning_rate": 7.773860182813403e-05,
+      "loss": 0.24,
+      "step": 704
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4634648263454437,
+      "learning_rate": 7.767469298995813e-05,
+      "loss": 0.3147,
+      "step": 705
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.44062742590904236,
+      "learning_rate": 7.761071890317994e-05,
+      "loss": 0.3168,
+      "step": 706
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.48678120970726013,
+      "learning_rate": 7.754667971863112e-05,
+      "loss": 0.3539,
+      "step": 707
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4622213542461395,
+      "learning_rate": 7.748257558729677e-05,
+      "loss": 0.4053,
+      "step": 708
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.47446855902671814,
+      "learning_rate": 7.741840666031517e-05,
+      "loss": 0.3993,
+      "step": 709
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4620521068572998,
+      "learning_rate": 7.735417308897736e-05,
+      "loss": 0.3956,
+      "step": 710
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.4270428717136383,
+      "learning_rate": 7.728987502472677e-05,
+      "loss": 0.2917,
+      "step": 711
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.42561075091362,
+      "learning_rate": 7.72255126191589e-05,
+      "loss": 0.3147,
+      "step": 712
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4774174392223358,
+      "learning_rate": 7.716108602402094e-05,
+      "loss": 0.377,
+      "step": 713
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4534321129322052,
+      "learning_rate": 7.709659539121144e-05,
+      "loss": 0.2768,
+      "step": 714
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.44026923179626465,
+      "learning_rate": 7.703204087277988e-05,
+      "loss": 0.3,
+      "step": 715
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.45911020040512085,
+      "learning_rate": 7.696742262092644e-05,
+      "loss": 0.3574,
+      "step": 716
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4503299295902252,
+      "learning_rate": 7.690274078800149e-05,
+      "loss": 0.323,
+      "step": 717
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4195705056190491,
+      "learning_rate": 7.683799552650534e-05,
+      "loss": 0.3255,
+      "step": 718
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.28857332468032837,
+      "learning_rate": 7.677318698908789e-05,
+      "loss": 0.1524,
+      "step": 719
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4472303092479706,
+      "learning_rate": 7.67083153285481e-05,
+      "loss": 0.2762,
+      "step": 720
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.4145956039428711,
+      "learning_rate": 7.66433806978339e-05,
+      "loss": 0.2748,
+      "step": 721
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.47666627168655396,
+      "learning_rate": 7.657838325004159e-05,
+      "loss": 0.3254,
+      "step": 722
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.37283122539520264,
+      "learning_rate": 7.651332313841562e-05,
+      "loss": 0.2643,
+      "step": 723
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.49776697158813477,
+      "learning_rate": 7.644820051634812e-05,
+      "loss": 0.3536,
+      "step": 724
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4178147614002228,
+      "learning_rate": 7.638301553737869e-05,
+      "loss": 0.3083,
+      "step": 725
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5427234172821045,
+      "learning_rate": 7.63177683551939e-05,
+      "loss": 0.3991,
+      "step": 726
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.42207062244415283,
+      "learning_rate": 7.625245912362699e-05,
+      "loss": 0.3213,
+      "step": 727
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4206954836845398,
+      "learning_rate": 7.618708799665744e-05,
+      "loss": 0.272,
+      "step": 728
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3957071900367737,
+      "learning_rate": 7.612165512841075e-05,
+      "loss": 0.3111,
+      "step": 729
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3757101595401764,
+      "learning_rate": 7.605616067315793e-05,
+      "loss": 0.2445,
+      "step": 730
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.4205314517021179,
+      "learning_rate": 7.59906047853152e-05,
+      "loss": 0.3324,
+      "step": 731
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.44776806235313416,
+      "learning_rate": 7.592498761944362e-05,
+      "loss": 0.3108,
+      "step": 732
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.42672890424728394,
+      "learning_rate": 7.585930933024875e-05,
+      "loss": 0.2884,
+      "step": 733
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3732846975326538,
+      "learning_rate": 7.579357007258022e-05,
+      "loss": 0.2664,
+      "step": 734
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.42130428552627563,
+      "learning_rate": 7.572777000143145e-05,
+      "loss": 0.329,
+      "step": 735
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.40204524993896484,
+      "learning_rate": 7.56619092719392e-05,
+      "loss": 0.2738,
+      "step": 736
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4448773264884949,
+      "learning_rate": 7.559598803938328e-05,
+      "loss": 0.3012,
+      "step": 737
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.41744062304496765,
+      "learning_rate": 7.553000645918611e-05,
+      "loss": 0.3414,
+      "step": 738
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.39561623334884644,
+      "learning_rate": 7.546396468691242e-05,
+      "loss": 0.2307,
+      "step": 739
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.39662009477615356,
+      "learning_rate": 7.539786287826885e-05,
+      "loss": 0.3337,
+      "step": 740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.43805012106895447,
+      "learning_rate": 7.533170118910355e-05,
+      "loss": 0.3815,
+      "step": 741
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4056797921657562,
+      "learning_rate": 7.526547977540592e-05,
+      "loss": 0.2854,
+      "step": 742
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3997124433517456,
+      "learning_rate": 7.519919879330614e-05,
+      "loss": 0.2793,
+      "step": 743
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.4407881498336792,
+      "learning_rate": 7.51328583990748e-05,
+      "loss": 0.3413,
+      "step": 744
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.41931819915771484,
+      "learning_rate": 7.506645874912264e-05,
+      "loss": 0.3285,
+      "step": 745
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4308168292045593,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.339,
+      "step": 746
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3031938672065735,
+      "learning_rate": 7.493348230839668e-05,
+      "loss": 0.2291,
+      "step": 747
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2818709909915924,
+      "learning_rate": 7.486690583114136e-05,
+      "loss": 0.1653,
+      "step": 748
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4321408271789551,
+      "learning_rate": 7.480027072520137e-05,
+      "loss": 0.3195,
+      "step": 749
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3854350745677948,
+      "learning_rate": 7.473357714768222e-05,
+      "loss": 0.3132,
+      "step": 750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4337480664253235,
+      "learning_rate": 7.46668252558273e-05,
+      "loss": 0.4228,
+      "step": 751
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.32639411091804504,
+      "learning_rate": 7.460001520701755e-05,
+      "loss": 0.22,
+      "step": 752
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.4251742660999298,
+      "learning_rate": 7.453314715877093e-05,
+      "loss": 0.4127,
+      "step": 753
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3949459493160248,
+      "learning_rate": 7.446622126874218e-05,
+      "loss": 0.2843,
+      "step": 754
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4311615526676178,
+      "learning_rate": 7.439923769472243e-05,
+      "loss": 0.321,
+      "step": 755
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.45880523324012756,
+      "learning_rate": 7.433219659463881e-05,
+      "loss": 0.364,
+      "step": 756
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3733888268470764,
+      "learning_rate": 7.426509812655406e-05,
+      "loss": 0.3111,
+      "step": 757
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3711423873901367,
+      "learning_rate": 7.41979424486662e-05,
+      "loss": 0.2316,
+      "step": 758
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.419147253036499,
+      "learning_rate": 7.413072971930807e-05,
+      "loss": 0.3635,
+      "step": 759
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.38382333517074585,
+      "learning_rate": 7.406346009694713e-05,
+      "loss": 0.2651,
+      "step": 760
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3497174382209778,
+      "learning_rate": 7.399613374018489e-05,
+      "loss": 0.245,
+      "step": 761
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.39361149072647095,
+      "learning_rate": 7.392875080775665e-05,
+      "loss": 0.2877,
+      "step": 762
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4261232316493988,
+      "learning_rate": 7.386131145853112e-05,
+      "loss": 0.3182,
+      "step": 763
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.40929925441741943,
+      "learning_rate": 7.379381585150996e-05,
+      "loss": 0.2263,
+      "step": 764
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.42151281237602234,
+      "learning_rate": 7.372626414582754e-05,
+      "loss": 0.3133,
+      "step": 765
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4434579610824585,
+      "learning_rate": 7.365865650075046e-05,
+      "loss": 0.3122,
+      "step": 766
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4639989137649536,
+      "learning_rate": 7.359099307567721e-05,
+      "loss": 0.3978,
+      "step": 767
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.37903618812561035,
+      "learning_rate": 7.35232740301378e-05,
+      "loss": 0.223,
+      "step": 768
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4896082580089569,
+      "learning_rate": 7.345549952379334e-05,
+      "loss": 0.298,
+      "step": 769
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.425721675157547,
+      "learning_rate": 7.338766971643579e-05,
+      "loss": 0.3022,
+      "step": 770
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4889935255050659,
+      "learning_rate": 7.331978476798738e-05,
+      "loss": 0.3128,
+      "step": 771
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.36715051531791687,
+      "learning_rate": 7.325184483850042e-05,
+      "loss": 0.2176,
+      "step": 772
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.41864633560180664,
+      "learning_rate": 7.318385008815686e-05,
+      "loss": 0.3162,
+      "step": 773
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4161873459815979,
+      "learning_rate": 7.311580067726783e-05,
+      "loss": 0.2691,
+      "step": 774
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.4776516556739807,
+      "learning_rate": 7.304769676627339e-05,
+      "loss": 0.3893,
+      "step": 775
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.43818777799606323,
+      "learning_rate": 7.297953851574206e-05,
+      "loss": 0.3441,
+      "step": 776
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.38630223274230957,
+      "learning_rate": 7.291132608637052e-05,
+      "loss": 0.2874,
+      "step": 777
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.33681949973106384,
+      "learning_rate": 7.284305963898314e-05,
+      "loss": 0.1887,
+      "step": 778
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4542236626148224,
+      "learning_rate": 7.27747393345317e-05,
+      "loss": 0.3674,
+      "step": 779
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.46607330441474915,
+      "learning_rate": 7.27063653340949e-05,
+      "loss": 0.3663,
+      "step": 780
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.37286949157714844,
+      "learning_rate": 7.263793779887808e-05,
+      "loss": 0.2737,
+      "step": 781
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5284208655357361,
+      "learning_rate": 7.25694568902128e-05,
+      "loss": 0.4026,
+      "step": 782
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4396823048591614,
+      "learning_rate": 7.250092276955642e-05,
+      "loss": 0.2903,
+      "step": 783
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.42539432644844055,
+      "learning_rate": 7.243233559849179e-05,
+      "loss": 0.2638,
+      "step": 784
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.396041601896286,
+      "learning_rate": 7.236369553872685e-05,
+      "loss": 0.3145,
+      "step": 785
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.38591712713241577,
+      "learning_rate": 7.229500275209418e-05,
+      "loss": 0.2877,
+      "step": 786
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4584382474422455,
+      "learning_rate": 7.222625740055073e-05,
+      "loss": 0.3889,
+      "step": 787
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.41197261214256287,
+      "learning_rate": 7.215745964617737e-05,
+      "loss": 0.2644,
+      "step": 788
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4837060570716858,
+      "learning_rate": 7.20886096511785e-05,
+      "loss": 0.3432,
+      "step": 789
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.49443215131759644,
+      "learning_rate": 7.201970757788172e-05,
+      "loss": 0.3229,
+      "step": 790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3780457377433777,
+      "learning_rate": 7.195075358873738e-05,
+      "loss": 0.2682,
+      "step": 791
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4492763876914978,
+      "learning_rate": 7.188174784631824e-05,
+      "loss": 0.3838,
+      "step": 792
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4437967836856842,
+      "learning_rate": 7.181269051331909e-05,
+      "loss": 0.3741,
+      "step": 793
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4623168110847473,
+      "learning_rate": 7.174358175255636e-05,
+      "loss": 0.306,
+      "step": 794
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4547882378101349,
+      "learning_rate": 7.16744217269677e-05,
+      "loss": 0.3374,
+      "step": 795
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.4731157720088959,
+      "learning_rate": 7.160521059961169e-05,
+      "loss": 0.3497,
+      "step": 796
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.45307835936546326,
+      "learning_rate": 7.153594853366731e-05,
+      "loss": 0.3008,
+      "step": 797
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.37926653027534485,
+      "learning_rate": 7.146663569243371e-05,
+      "loss": 0.2632,
+      "step": 798
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3968765437602997,
+      "learning_rate": 7.139727223932968e-05,
+      "loss": 0.2296,
+      "step": 799
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.36611175537109375,
+      "learning_rate": 7.132785833789344e-05,
+      "loss": 0.2093,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.29461246728897095,
+      "eval_runtime": 378.9835,
+      "eval_samples_per_second": 0.9,
+      "eval_steps_per_second": 0.227,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3777669370174408,
+      "learning_rate": 7.125839415178204e-05,
+      "loss": 0.2181,
+      "step": 801
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5459440350532532,
+      "learning_rate": 7.118887984477116e-05,
+      "loss": 0.4904,
+      "step": 802
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3982379138469696,
+      "learning_rate": 7.111931558075464e-05,
+      "loss": 0.3064,
+      "step": 803
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4211267828941345,
+      "learning_rate": 7.104970152374405e-05,
+      "loss": 0.2904,
+      "step": 804
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.41220736503601074,
+      "learning_rate": 7.098003783786845e-05,
+      "loss": 0.2048,
+      "step": 805
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.40161508321762085,
+      "learning_rate": 7.091032468737381e-05,
+      "loss": 0.2226,
+      "step": 806
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3855397701263428,
+      "learning_rate": 7.084056223662282e-05,
+      "loss": 0.2498,
+      "step": 807
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3812676668167114,
+      "learning_rate": 7.077075065009433e-05,
+      "loss": 0.2239,
+      "step": 808
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4389495849609375,
+      "learning_rate": 7.070089009238306e-05,
+      "loss": 0.3442,
+      "step": 809
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.38800790905952454,
+      "learning_rate": 7.063098072819919e-05,
+      "loss": 0.2597,
+      "step": 810
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3808739185333252,
+      "learning_rate": 7.056102272236799e-05,
+      "loss": 0.2926,
+      "step": 811
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4258776903152466,
+      "learning_rate": 7.049101623982937e-05,
+      "loss": 0.2858,
+      "step": 812
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4349055886268616,
+      "learning_rate": 7.04209614456376e-05,
+      "loss": 0.3234,
+      "step": 813
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4832972586154938,
+      "learning_rate": 7.03508585049608e-05,
+      "loss": 0.3225,
+      "step": 814
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.4321622848510742,
+      "learning_rate": 7.028070758308059e-05,
+      "loss": 0.3285,
+      "step": 815
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3984975516796112,
+      "learning_rate": 7.021050884539177e-05,
+      "loss": 0.3053,
+      "step": 816
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5002953410148621,
+      "learning_rate": 7.014026245740186e-05,
+      "loss": 0.3253,
+      "step": 817
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4297904074192047,
+      "learning_rate": 7.006996858473069e-05,
+      "loss": 0.3438,
+      "step": 818
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.40808209776878357,
+      "learning_rate": 6.999962739311008e-05,
+      "loss": 0.2601,
+      "step": 819
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3974955677986145,
+      "learning_rate": 6.99292390483834e-05,
+      "loss": 0.2646,
+      "step": 820
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4036179482936859,
+      "learning_rate": 6.98588037165052e-05,
+      "loss": 0.2748,
+      "step": 821
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.43644773960113525,
+      "learning_rate": 6.97883215635408e-05,
+      "loss": 0.4035,
+      "step": 822
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.4592438042163849,
+      "learning_rate": 6.971779275566593e-05,
+      "loss": 0.3097,
+      "step": 823
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.44867798686027527,
+      "learning_rate": 6.964721745916631e-05,
+      "loss": 0.3469,
+      "step": 824
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3540700376033783,
+      "learning_rate": 6.957659584043724e-05,
+      "loss": 0.2409,
+      "step": 825
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.39834967255592346,
+      "learning_rate": 6.950592806598327e-05,
+      "loss": 0.2836,
+      "step": 826
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.42365822196006775,
+      "learning_rate": 6.943521430241776e-05,
+      "loss": 0.3609,
+      "step": 827
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.35388749837875366,
+      "learning_rate": 6.936445471646249e-05,
+      "loss": 0.2388,
+      "step": 828
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.42534494400024414,
+      "learning_rate": 6.929364947494728e-05,
+      "loss": 0.364,
+      "step": 829
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.49986061453819275,
+      "learning_rate": 6.922279874480959e-05,
+      "loss": 0.3876,
+      "step": 830
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.45167818665504456,
+      "learning_rate": 6.915190269309414e-05,
+      "loss": 0.4129,
+      "step": 831
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5426003336906433,
+      "learning_rate": 6.908096148695251e-05,
+      "loss": 0.2874,
+      "step": 832
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.34209468960762024,
+      "learning_rate": 6.90099752936427e-05,
+      "loss": 0.1849,
+      "step": 833
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4010191857814789,
+      "learning_rate": 6.89389442805288e-05,
+      "loss": 0.2937,
+      "step": 834
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.43319109082221985,
+      "learning_rate": 6.886786861508061e-05,
+      "loss": 0.3331,
+      "step": 835
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4459756016731262,
+      "learning_rate": 6.879674846487314e-05,
+      "loss": 0.3244,
+      "step": 836
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.478600412607193,
+      "learning_rate": 6.872558399758632e-05,
+      "loss": 0.3612,
+      "step": 837
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.41877803206443787,
+      "learning_rate": 6.865437538100457e-05,
+      "loss": 0.34,
+      "step": 838
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.4314415752887726,
+      "learning_rate": 6.858312278301637e-05,
+      "loss": 0.367,
+      "step": 839
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.48531678318977356,
+      "learning_rate": 6.851182637161396e-05,
+      "loss": 0.3783,
+      "step": 840
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.48599305748939514,
+      "learning_rate": 6.844048631489278e-05,
+      "loss": 0.3563,
+      "step": 841
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4623684585094452,
+      "learning_rate": 6.836910278105124e-05,
+      "loss": 0.2707,
+      "step": 842
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.38649484515190125,
+      "learning_rate": 6.829767593839028e-05,
+      "loss": 0.263,
+      "step": 843
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4263112246990204,
+      "learning_rate": 6.822620595531286e-05,
+      "loss": 0.3161,
+      "step": 844
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4235820770263672,
+      "learning_rate": 6.815469300032374e-05,
+      "loss": 0.3044,
+      "step": 845
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4437580108642578,
+      "learning_rate": 6.808313724202894e-05,
+      "loss": 0.403,
+      "step": 846
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3393166661262512,
+      "learning_rate": 6.801153884913542e-05,
+      "loss": 0.2466,
+      "step": 847
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4446983337402344,
+      "learning_rate": 6.793989799045066e-05,
+      "loss": 0.2929,
+      "step": 848
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4104369580745697,
+      "learning_rate": 6.786821483488226e-05,
+      "loss": 0.2429,
+      "step": 849
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.560243546962738,
+      "learning_rate": 6.779648955143754e-05,
+      "loss": 0.4037,
+      "step": 850
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4177003800868988,
+      "learning_rate": 6.772472230922313e-05,
+      "loss": 0.3197,
+      "step": 851
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4375431537628174,
+      "learning_rate": 6.765291327744463e-05,
+      "loss": 0.3026,
+      "step": 852
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3986719846725464,
+      "learning_rate": 6.758106262540611e-05,
+      "loss": 0.2631,
+      "step": 853
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3859230577945709,
+      "learning_rate": 6.750917052250982e-05,
+      "loss": 0.2697,
+      "step": 854
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3835771381855011,
+      "learning_rate": 6.743723713825569e-05,
+      "loss": 0.3371,
+      "step": 855
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3162922263145447,
+      "learning_rate": 6.736526264224101e-05,
+      "loss": 0.2569,
+      "step": 856
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4173866808414459,
+      "learning_rate": 6.729324720416002e-05,
+      "loss": 0.2862,
+      "step": 857
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2804356813430786,
+      "learning_rate": 6.722119099380345e-05,
+      "loss": 0.1369,
+      "step": 858
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3329291045665741,
+      "learning_rate": 6.714909418105815e-05,
+      "loss": 0.2348,
+      "step": 859
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.4261725842952728,
+      "learning_rate": 6.707695693590676e-05,
+      "loss": 0.4068,
+      "step": 860
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3827839493751526,
+      "learning_rate": 6.700477942842717e-05,
+      "loss": 0.2789,
+      "step": 861
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.44311416149139404,
+      "learning_rate": 6.693256182879225e-05,
+      "loss": 0.3463,
+      "step": 862
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.33683061599731445,
+      "learning_rate": 6.686030430726937e-05,
+      "loss": 0.191,
+      "step": 863
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.37077856063842773,
+      "learning_rate": 6.678800703422003e-05,
+      "loss": 0.2843,
+      "step": 864
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.4832366406917572,
+      "learning_rate": 6.671567018009947e-05,
+      "loss": 0.3528,
+      "step": 865
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.382205992937088,
+      "learning_rate": 6.664329391545626e-05,
+      "loss": 0.2291,
+      "step": 866
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.38632795214653015,
+      "learning_rate": 6.65708784109318e-05,
+      "loss": 0.2668,
+      "step": 867
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3762361407279968,
+      "learning_rate": 6.64984238372601e-05,
+      "loss": 0.2693,
+      "step": 868
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.41171541810035706,
+      "learning_rate": 6.642593036526728e-05,
+      "loss": 0.2652,
+      "step": 869
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3439303934574127,
+      "learning_rate": 6.635339816587109e-05,
+      "loss": 0.1714,
+      "step": 870
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.40999844670295715,
+      "learning_rate": 6.628082741008069e-05,
+      "loss": 0.327,
+      "step": 871
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4489401876926422,
+      "learning_rate": 6.620821826899606e-05,
+      "loss": 0.3741,
+      "step": 872
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3974386751651764,
+      "learning_rate": 6.613557091380771e-05,
+      "loss": 0.2644,
+      "step": 873
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.40071815252304077,
+      "learning_rate": 6.606288551579629e-05,
+      "loss": 0.2836,
+      "step": 874
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4391033947467804,
+      "learning_rate": 6.599016224633209e-05,
+      "loss": 0.3373,
+      "step": 875
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.38056957721710205,
+      "learning_rate": 6.59174012768747e-05,
+      "loss": 0.1978,
+      "step": 876
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.39703795313835144,
+      "learning_rate": 6.584460277897263e-05,
+      "loss": 0.2643,
+      "step": 877
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.35327187180519104,
+      "learning_rate": 6.577176692426279e-05,
+      "loss": 0.2151,
+      "step": 878
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.4626995623111725,
+      "learning_rate": 6.569889388447025e-05,
+      "loss": 0.3679,
+      "step": 879
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5043691992759705,
+      "learning_rate": 6.562598383140772e-05,
+      "loss": 0.3788,
+      "step": 880
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6156384348869324,
+      "learning_rate": 6.555303693697517e-05,
+      "loss": 0.4508,
+      "step": 881
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4287849962711334,
+      "learning_rate": 6.548005337315944e-05,
+      "loss": 0.3515,
+      "step": 882
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4149426519870758,
+      "learning_rate": 6.540703331203381e-05,
+      "loss": 0.2986,
+      "step": 883
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.46264752745628357,
+      "learning_rate": 6.533397692575766e-05,
+      "loss": 0.4091,
+      "step": 884
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3814181089401245,
+      "learning_rate": 6.526088438657594e-05,
+      "loss": 0.2629,
+      "step": 885
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.340842604637146,
+      "learning_rate": 6.518775586681887e-05,
+      "loss": 0.2071,
+      "step": 886
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4531356990337372,
+      "learning_rate": 6.511459153890152e-05,
+      "loss": 0.3194,
+      "step": 887
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4712303876876831,
+      "learning_rate": 6.504139157532337e-05,
+      "loss": 0.3165,
+      "step": 888
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.44127264618873596,
+      "learning_rate": 6.496815614866791e-05,
+      "loss": 0.3226,
+      "step": 889
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.4647597372531891,
+      "learning_rate": 6.489488543160224e-05,
+      "loss": 0.3399,
+      "step": 890
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.41681021451950073,
+      "learning_rate": 6.48215795968767e-05,
+      "loss": 0.3032,
+      "step": 891
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3830890357494354,
+      "learning_rate": 6.47482388173244e-05,
+      "loss": 0.2655,
+      "step": 892
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40932777523994446,
+      "learning_rate": 6.46748632658608e-05,
+      "loss": 0.2958,
+      "step": 893
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4666089713573456,
+      "learning_rate": 6.460145311548341e-05,
+      "loss": 0.3443,
+      "step": 894
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.48490193486213684,
+      "learning_rate": 6.452800853927128e-05,
+      "loss": 0.3166,
+      "step": 895
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.506223738193512,
+      "learning_rate": 6.445452971038464e-05,
+      "loss": 0.4035,
+      "step": 896
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3868057131767273,
+      "learning_rate": 6.438101680206444e-05,
+      "loss": 0.2299,
+      "step": 897
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4486956298351288,
+      "learning_rate": 6.430746998763204e-05,
+      "loss": 0.3622,
+      "step": 898
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.4172947108745575,
+      "learning_rate": 6.42338894404887e-05,
+      "loss": 0.3228,
+      "step": 899
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3854050934314728,
+      "learning_rate": 6.416027533411519e-05,
+      "loss": 0.295,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.28741392493247986,
+      "eval_runtime": 373.4695,
+      "eval_samples_per_second": 0.913,
+      "eval_steps_per_second": 0.23,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.44058945775032043,
+      "learning_rate": 6.408662784207149e-05,
+      "loss": 0.3463,
+      "step": 901
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3871374726295471,
+      "learning_rate": 6.401294713799618e-05,
+      "loss": 0.3158,
+      "step": 902
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3712465465068817,
+      "learning_rate": 6.393923339560621e-05,
+      "loss": 0.308,
+      "step": 903
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.407106876373291,
+      "learning_rate": 6.386548678869644e-05,
+      "loss": 0.3006,
+      "step": 904
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.41011539101600647,
+      "learning_rate": 6.379170749113917e-05,
+      "loss": 0.3241,
+      "step": 905
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.37096431851387024,
+      "learning_rate": 6.37178956768838e-05,
+      "loss": 0.2348,
+      "step": 906
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.38033682107925415,
+      "learning_rate": 6.364405151995635e-05,
+      "loss": 0.3181,
+      "step": 907
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3767927885055542,
+      "learning_rate": 6.35701751944592e-05,
+      "loss": 0.287,
+      "step": 908
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4490831792354584,
+      "learning_rate": 6.349626687457045e-05,
+      "loss": 0.3335,
+      "step": 909
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.4378523528575897,
+      "learning_rate": 6.342232673454371e-05,
+      "loss": 0.3324,
+      "step": 910
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.35792258381843567,
+      "learning_rate": 6.334835494870758e-05,
+      "loss": 0.2233,
+      "step": 911
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3392629623413086,
+      "learning_rate": 6.32743516914653e-05,
+      "loss": 0.2203,
+      "step": 912
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.36681029200553894,
+      "learning_rate": 6.320031713729428e-05,
+      "loss": 0.2987,
+      "step": 913
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.38458046317100525,
+      "learning_rate": 6.312625146074575e-05,
+      "loss": 0.2753,
+      "step": 914
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.39717918634414673,
+      "learning_rate": 6.305215483644426e-05,
+      "loss": 0.2363,
+      "step": 915
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3940262496471405,
+      "learning_rate": 6.297802743908741e-05,
+      "loss": 0.2245,
+      "step": 916
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.35425087809562683,
+      "learning_rate": 6.290386944344528e-05,
+      "loss": 0.2197,
+      "step": 917
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4860681891441345,
+      "learning_rate": 6.282968102436009e-05,
+      "loss": 0.3614,
+      "step": 918
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.4314277768135071,
+      "learning_rate": 6.275546235674588e-05,
+      "loss": 0.2838,
+      "step": 919
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.44403505325317383,
+      "learning_rate": 6.268121361558792e-05,
+      "loss": 0.3162,
+      "step": 920
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5254203677177429,
+      "learning_rate": 6.260693497594242e-05,
+      "loss": 0.3472,
+      "step": 921
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.47920742630958557,
+      "learning_rate": 6.253262661293604e-05,
+      "loss": 0.3358,
+      "step": 922
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.38567379117012024,
+      "learning_rate": 6.245828870176557e-05,
+      "loss": 0.2288,
+      "step": 923
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.42042529582977295,
+      "learning_rate": 6.238392141769743e-05,
+      "loss": 0.2524,
+      "step": 924
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.46694624423980713,
+      "learning_rate": 6.230952493606734e-05,
+      "loss": 0.3253,
+      "step": 925
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.41947153210639954,
+      "learning_rate": 6.22350994322798e-05,
+      "loss": 0.3286,
+      "step": 926
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.46464383602142334,
+      "learning_rate": 6.216064508180778e-05,
+      "loss": 0.3734,
+      "step": 927
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4532264769077301,
+      "learning_rate": 6.208616206019224e-05,
+      "loss": 0.3114,
+      "step": 928
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4733847677707672,
+      "learning_rate": 6.201165054304173e-05,
+      "loss": 0.3186,
+      "step": 929
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4441421926021576,
+      "learning_rate": 6.193711070603202e-05,
+      "loss": 0.3232,
+      "step": 930
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4824698269367218,
+      "learning_rate": 6.18625427249056e-05,
+      "loss": 0.3027,
+      "step": 931
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4747307300567627,
+      "learning_rate": 6.178794677547137e-05,
+      "loss": 0.4075,
+      "step": 932
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.36383411288261414,
+      "learning_rate": 6.171332303360411e-05,
+      "loss": 0.2394,
+      "step": 933
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.401227742433548,
+      "learning_rate": 6.163867167524419e-05,
+      "loss": 0.2381,
+      "step": 934
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4125315546989441,
+      "learning_rate": 6.156399287639703e-05,
+      "loss": 0.3283,
+      "step": 935
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41893908381462097,
+      "learning_rate": 6.14892868131328e-05,
+      "loss": 0.2625,
+      "step": 936
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.39285755157470703,
+      "learning_rate": 6.14145536615859e-05,
+      "loss": 0.2309,
+      "step": 937
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.39775630831718445,
+      "learning_rate": 6.133979359795468e-05,
+      "loss": 0.2204,
+      "step": 938
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41518449783325195,
+      "learning_rate": 6.126500679850083e-05,
+      "loss": 0.2981,
+      "step": 939
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.41295257210731506,
+      "learning_rate": 6.119019343954914e-05,
+      "loss": 0.3191,
+      "step": 940
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.40648797154426575,
+      "learning_rate": 6.111535369748702e-05,
+      "loss": 0.3048,
+      "step": 941
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5105196237564087,
+      "learning_rate": 6.104048774876406e-05,
+      "loss": 0.3862,
+      "step": 942
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4334690272808075,
+      "learning_rate": 6.096559576989166e-05,
+      "loss": 0.3477,
+      "step": 943
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.43452736735343933,
+      "learning_rate": 6.0890677937442574e-05,
+      "loss": 0.3336,
+      "step": 944
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.38207656145095825,
+      "learning_rate": 6.0815734428050533e-05,
+      "loss": 0.2916,
+      "step": 945
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4759387671947479,
+      "learning_rate": 6.074076541840977e-05,
+      "loss": 0.3275,
+      "step": 946
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.420587420463562,
+      "learning_rate": 6.066577108527468e-05,
+      "loss": 0.3348,
+      "step": 947
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4614073932170868,
+      "learning_rate": 6.059075160545933e-05,
+      "loss": 0.417,
+      "step": 948
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4970424473285675,
+      "learning_rate": 6.05157071558371e-05,
+      "loss": 0.3921,
+      "step": 949
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4004935324192047,
+      "learning_rate": 6.0440637913340224e-05,
+      "loss": 0.2701,
+      "step": 950
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.49780601263046265,
+      "learning_rate": 6.03655440549594e-05,
+      "loss": 0.3838,
+      "step": 951
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.4134318232536316,
+      "learning_rate": 6.029042575774334e-05,
+      "loss": 0.3306,
+      "step": 952
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5322731733322144,
+      "learning_rate": 6.0215283198798436e-05,
+      "loss": 0.4219,
+      "step": 953
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3864912688732147,
+      "learning_rate": 6.0140116555288204e-05,
+      "loss": 0.2959,
+      "step": 954
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.43973082304000854,
+      "learning_rate": 6.006492600443301e-05,
+      "loss": 0.3543,
+      "step": 955
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.35023313760757446,
+      "learning_rate": 5.998971172350953e-05,
+      "loss": 0.2122,
+      "step": 956
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3881041705608368,
+      "learning_rate": 5.991447388985045e-05,
+      "loss": 0.2606,
+      "step": 957
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.45033326745033264,
+      "learning_rate": 5.9839212680843925e-05,
+      "loss": 0.3712,
+      "step": 958
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4287780821323395,
+      "learning_rate": 5.9763928273933256e-05,
+      "loss": 0.3328,
+      "step": 959
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.40677621960639954,
+      "learning_rate": 5.968862084661643e-05,
+      "loss": 0.2965,
+      "step": 960
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.445709228515625,
+      "learning_rate": 5.9613290576445704e-05,
+      "loss": 0.3621,
+      "step": 961
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.46743589639663696,
+      "learning_rate": 5.953793764102722e-05,
+      "loss": 0.3045,
+      "step": 962
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.43945255875587463,
+      "learning_rate": 5.946256221802051e-05,
+      "loss": 0.3155,
+      "step": 963
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.43651899695396423,
+      "learning_rate": 5.938716448513818e-05,
+      "loss": 0.3273,
+      "step": 964
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4564177095890045,
+      "learning_rate": 5.931174462014537e-05,
+      "loss": 0.3897,
+      "step": 965
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4437497854232788,
+      "learning_rate": 5.923630280085948e-05,
+      "loss": 0.314,
+      "step": 966
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.49537619948387146,
+      "learning_rate": 5.916083920514959e-05,
+      "loss": 0.2921,
+      "step": 967
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.353377103805542,
+      "learning_rate": 5.908535401093618e-05,
+      "loss": 0.2571,
+      "step": 968
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.44651278853416443,
+      "learning_rate": 5.900984739619062e-05,
+      "loss": 0.3335,
+      "step": 969
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.35381266474723816,
+      "learning_rate": 5.8934319538934824e-05,
+      "loss": 0.2132,
+      "step": 970
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.37143078446388245,
+      "learning_rate": 5.8858770617240746e-05,
+      "loss": 0.265,
+      "step": 971
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4304827153682709,
+      "learning_rate": 5.878320080923001e-05,
+      "loss": 0.3441,
+      "step": 972
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.4664653241634369,
+      "learning_rate": 5.870761029307352e-05,
+      "loss": 0.2296,
+      "step": 973
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.587513267993927,
+      "learning_rate": 5.863199924699095e-05,
+      "loss": 0.5095,
+      "step": 974
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.45028263330459595,
+      "learning_rate": 5.8556367849250436e-05,
+      "loss": 0.3131,
+      "step": 975
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3657177686691284,
+      "learning_rate": 5.848071627816803e-05,
+      "loss": 0.2022,
+      "step": 976
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.37873032689094543,
+      "learning_rate": 5.840504471210742e-05,
+      "loss": 0.2389,
+      "step": 977
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3741288185119629,
+      "learning_rate": 5.8329353329479366e-05,
+      "loss": 0.266,
+      "step": 978
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.45317283272743225,
+      "learning_rate": 5.82536423087414e-05,
+      "loss": 0.3264,
+      "step": 979
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4137780964374542,
+      "learning_rate": 5.8177911828397336e-05,
+      "loss": 0.2847,
+      "step": 980
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3918278217315674,
+      "learning_rate": 5.8102162066996854e-05,
+      "loss": 0.3141,
+      "step": 981
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.47474202513694763,
+      "learning_rate": 5.802639320313514e-05,
+      "loss": 0.4333,
+      "step": 982
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3890018165111542,
+      "learning_rate": 5.795060541545236e-05,
+      "loss": 0.2969,
+      "step": 983
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.40383580327033997,
+      "learning_rate": 5.787479888263333e-05,
+      "loss": 0.3061,
+      "step": 984
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4334053099155426,
+      "learning_rate": 5.779897378340705e-05,
+      "loss": 0.3007,
+      "step": 985
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3814491927623749,
+      "learning_rate": 5.77231302965463e-05,
+      "loss": 0.2928,
+      "step": 986
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3657767176628113,
+      "learning_rate": 5.76472686008672e-05,
+      "loss": 0.2676,
+      "step": 987
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.34177902340888977,
+      "learning_rate": 5.757138887522884e-05,
+      "loss": 0.2219,
+      "step": 988
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.48286789655685425,
+      "learning_rate": 5.7495491298532765e-05,
+      "loss": 0.4197,
+      "step": 989
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4207167327404022,
+      "learning_rate": 5.7419576049722646e-05,
+      "loss": 0.2855,
+      "step": 990
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3679610788822174,
+      "learning_rate": 5.73436433077838e-05,
+      "loss": 0.2168,
+      "step": 991
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.38789647817611694,
+      "learning_rate": 5.7267693251742794e-05,
+      "loss": 0.2327,
+      "step": 992
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.41098183393478394,
+      "learning_rate": 5.719172606066703e-05,
+      "loss": 0.3268,
+      "step": 993
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.49737051129341125,
+      "learning_rate": 5.7115741913664264e-05,
+      "loss": 0.4083,
+      "step": 994
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.4628293812274933,
+      "learning_rate": 5.703974098988229e-05,
+      "loss": 0.328,
+      "step": 995
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3253394663333893,
+      "learning_rate": 5.696372346850841e-05,
+      "loss": 0.2148,
+      "step": 996
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.42477867007255554,
+      "learning_rate": 5.68876895287691e-05,
+      "loss": 0.2954,
+      "step": 997
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4249178171157837,
+      "learning_rate": 5.68116393499295e-05,
+      "loss": 0.2415,
+      "step": 998
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3576670289039612,
+      "learning_rate": 5.673557311129306e-05,
+      "loss": 0.2196,
+      "step": 999
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.436309814453125,
+      "learning_rate": 5.6659490992201094e-05,
+      "loss": 0.337,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.28351831436157227,
+      "eval_runtime": 374.8413,
+      "eval_samples_per_second": 0.91,
+      "eval_steps_per_second": 0.229,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.43124595284461975,
+      "learning_rate": 5.658339317203235e-05,
+      "loss": 0.2578,
+      "step": 1001
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3619249761104584,
+      "learning_rate": 5.650727983020262e-05,
+      "loss": 0.2087,
+      "step": 1002
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.41123858094215393,
+      "learning_rate": 5.643115114616425e-05,
+      "loss": 0.3175,
+      "step": 1003
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.4604555368423462,
+      "learning_rate": 5.635500729940577e-05,
+      "loss": 0.3067,
+      "step": 1004
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.44970881938934326,
+      "learning_rate": 5.627884846945151e-05,
+      "loss": 0.2836,
+      "step": 1005
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.45657435059547424,
+      "learning_rate": 5.620267483586105e-05,
+      "loss": 0.2654,
+      "step": 1006
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5313488245010376,
+      "learning_rate": 5.612648657822892e-05,
+      "loss": 0.3791,
+      "step": 1007
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4581022262573242,
+      "learning_rate": 5.6050283876184116e-05,
+      "loss": 0.2633,
+      "step": 1008
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3563360869884491,
+      "learning_rate": 5.597406690938969e-05,
+      "loss": 0.1823,
+      "step": 1009
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.4619802236557007,
+      "learning_rate": 5.5897835857542317e-05,
+      "loss": 0.2543,
+      "step": 1010
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.42287588119506836,
+      "learning_rate": 5.5821590900371886e-05,
+      "loss": 0.3411,
+      "step": 1011
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.43619757890701294,
+      "learning_rate": 5.574533221764109e-05,
+      "loss": 0.4003,
+      "step": 1012
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.48663079738616943,
+      "learning_rate": 5.566905998914496e-05,
+      "loss": 0.3868,
+      "step": 1013
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5171478986740112,
+      "learning_rate": 5.5592774394710465e-05,
+      "loss": 0.3593,
+      "step": 1014
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3793003261089325,
+      "learning_rate": 5.551647561419611e-05,
+      "loss": 0.258,
+      "step": 1015
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3943827450275421,
+      "learning_rate": 5.5440163827491454e-05,
+      "loss": 0.2649,
+      "step": 1016
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.37953877449035645,
+      "learning_rate": 5.536383921451673e-05,
+      "loss": 0.2638,
+      "step": 1017
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.42410457134246826,
+      "learning_rate": 5.5287501955222444e-05,
+      "loss": 0.2899,
+      "step": 1018
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.49995243549346924,
+      "learning_rate": 5.521115222958888e-05,
+      "loss": 0.3907,
+      "step": 1019
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.43869611620903015,
+      "learning_rate": 5.513479021762572e-05,
+      "loss": 0.316,
+      "step": 1020
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4236672818660736,
+      "learning_rate": 5.505841609937161e-05,
+      "loss": 0.3367,
+      "step": 1021
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.41587209701538086,
+      "learning_rate": 5.4982030054893785e-05,
+      "loss": 0.3263,
+      "step": 1022
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4228578209877014,
+      "learning_rate": 5.490563226428755e-05,
+      "loss": 0.2477,
+      "step": 1023
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.4622528553009033,
+      "learning_rate": 5.482922290767589e-05,
+      "loss": 0.3822,
+      "step": 1024
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.34737443923950195,
+      "learning_rate": 5.475280216520913e-05,
+      "loss": 0.2422,
+      "step": 1025
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.41179901361465454,
+      "learning_rate": 5.467637021706438e-05,
+      "loss": 0.2708,
+      "step": 1026
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3536500334739685,
+      "learning_rate": 5.4599927243445156e-05,
+      "loss": 0.2091,
+      "step": 1027
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.44380703568458557,
+      "learning_rate": 5.452347342458104e-05,
+      "loss": 0.3459,
+      "step": 1028
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.421606183052063,
+      "learning_rate": 5.4447008940727115e-05,
+      "loss": 0.3172,
+      "step": 1029
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5172922611236572,
+      "learning_rate": 5.437053397216364e-05,
+      "loss": 0.3716,
+      "step": 1030
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5004053711891174,
+      "learning_rate": 5.429404869919559e-05,
+      "loss": 0.4429,
+      "step": 1031
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.41400447487831116,
+      "learning_rate": 5.4217553302152237e-05,
+      "loss": 0.2896,
+      "step": 1032
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.42059293389320374,
+      "learning_rate": 5.414104796138673e-05,
+      "loss": 0.2883,
+      "step": 1033
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3976444602012634,
+      "learning_rate": 5.406453285727564e-05,
+      "loss": 0.2374,
+      "step": 1034
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.31789836287498474,
+      "learning_rate": 5.398800817021856e-05,
+      "loss": 0.1558,
+      "step": 1035
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4580276608467102,
+      "learning_rate": 5.39114740806377e-05,
+      "loss": 0.2791,
+      "step": 1036
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3810381293296814,
+      "learning_rate": 5.383493076897742e-05,
+      "loss": 0.2423,
+      "step": 1037
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4831598699092865,
+      "learning_rate": 5.3758378415703825e-05,
+      "loss": 0.3447,
+      "step": 1038
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.4247997999191284,
+      "learning_rate": 5.3681817201304344e-05,
+      "loss": 0.2876,
+      "step": 1039
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4443676471710205,
+      "learning_rate": 5.360524730628728e-05,
+      "loss": 0.2878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.39276984333992004,
+      "learning_rate": 5.352866891118142e-05,
+      "loss": 0.2751,
+      "step": 1041
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3604837954044342,
+      "learning_rate": 5.345208219653561e-05,
+      "loss": 0.2455,
+      "step": 1042
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3973501920700073,
+      "learning_rate": 5.337548734291826e-05,
+      "loss": 0.2578,
+      "step": 1043
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.457567036151886,
+      "learning_rate": 5.3298884530917014e-05,
+      "loss": 0.3434,
+      "step": 1044
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.35174161195755005,
+      "learning_rate": 5.322227394113825e-05,
+      "loss": 0.2031,
+      "step": 1045
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.4216848313808441,
+      "learning_rate": 5.3145655754206714e-05,
+      "loss": 0.2553,
+      "step": 1046
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.42642438411712646,
+      "learning_rate": 5.306903015076502e-05,
+      "loss": 0.2917,
+      "step": 1047
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3854960501194,
+      "learning_rate": 5.2992397311473316e-05,
+      "loss": 0.2464,
+      "step": 1048
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.428726464509964,
+      "learning_rate": 5.291575741700877e-05,
+      "loss": 0.3426,
+      "step": 1049
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.49436289072036743,
+      "learning_rate": 5.2839110648065214e-05,
+      "loss": 0.3796,
+      "step": 1050
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3638419806957245,
+      "learning_rate": 5.276245718535269e-05,
+      "loss": 0.2367,
+      "step": 1051
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.40477246046066284,
+      "learning_rate": 5.2685797209596976e-05,
+      "loss": 0.2564,
+      "step": 1052
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.44663867354393005,
+      "learning_rate": 5.260913090153927e-05,
+      "loss": 0.3713,
+      "step": 1053
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4775208532810211,
+      "learning_rate": 5.2532458441935636e-05,
+      "loss": 0.3259,
+      "step": 1054
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3852700889110565,
+      "learning_rate": 5.24557800115567e-05,
+      "loss": 0.2668,
+      "step": 1055
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4094158113002777,
+      "learning_rate": 5.2379095791187124e-05,
+      "loss": 0.3159,
+      "step": 1056
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3308325409889221,
+      "learning_rate": 5.2302405961625225e-05,
+      "loss": 0.1965,
+      "step": 1057
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4158184826374054,
+      "learning_rate": 5.222571070368257e-05,
+      "loss": 0.3068,
+      "step": 1058
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.32684066891670227,
+      "learning_rate": 5.214901019818352e-05,
+      "loss": 0.1649,
+      "step": 1059
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.33532243967056274,
+      "learning_rate": 5.2072304625964785e-05,
+      "loss": 0.1798,
+      "step": 1060
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.2880372405052185,
+      "learning_rate": 5.199559416787503e-05,
+      "loss": 0.1692,
+      "step": 1061
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3458213806152344,
+      "learning_rate": 5.191887900477443e-05,
+      "loss": 0.2351,
+      "step": 1062
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3412489891052246,
+      "learning_rate": 5.18421593175343e-05,
+      "loss": 0.188,
+      "step": 1063
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.32257041335105896,
+      "learning_rate": 5.176543528703657e-05,
+      "loss": 0.1752,
+      "step": 1064
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.4158090651035309,
+      "learning_rate": 5.168870709417342e-05,
+      "loss": 0.2474,
+      "step": 1065
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3195403516292572,
+      "learning_rate": 5.161197491984684e-05,
+      "loss": 0.1868,
+      "step": 1066
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3848859965801239,
+      "learning_rate": 5.153523894496826e-05,
+      "loss": 0.1827,
+      "step": 1067
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.4013334810733795,
+      "learning_rate": 5.1458499350458e-05,
+      "loss": 0.2655,
+      "step": 1068
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.4084205627441406,
+      "learning_rate": 5.1381756317244945e-05,
+      "loss": 0.2037,
+      "step": 1069
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.41331103444099426,
+      "learning_rate": 5.130501002626609e-05,
+      "loss": 0.2046,
+      "step": 1070
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3710659146308899,
+      "learning_rate": 5.1228260658466096e-05,
+      "loss": 0.1604,
+      "step": 1071
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4189535081386566,
+      "learning_rate": 5.11515083947969e-05,
+      "loss": 0.1814,
+      "step": 1072
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.47776368260383606,
+      "learning_rate": 5.107475341621726e-05,
+      "loss": 0.248,
+      "step": 1073
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4170460104942322,
+      "learning_rate": 5.099799590369231e-05,
+      "loss": 0.1671,
+      "step": 1074
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3788929879665375,
+      "learning_rate": 5.092123603819318e-05,
+      "loss": 0.2053,
+      "step": 1075
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4501115381717682,
+      "learning_rate": 5.084447400069655e-05,
+      "loss": 0.2409,
+      "step": 1076
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3974234461784363,
+      "learning_rate": 5.0767709972184244e-05,
+      "loss": 0.1959,
+      "step": 1077
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.40462905168533325,
+      "learning_rate": 5.069094413364272e-05,
+      "loss": 0.2099,
+      "step": 1078
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.48999881744384766,
+      "learning_rate": 5.061417666606274e-05,
+      "loss": 0.2179,
+      "step": 1079
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.35690560936927795,
+      "learning_rate": 5.053740775043892e-05,
+      "loss": 0.1609,
+      "step": 1080
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.3996225893497467,
+      "learning_rate": 5.046063756776925e-05,
+      "loss": 0.1487,
+      "step": 1081
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.45011118054389954,
+      "learning_rate": 5.038386629905475e-05,
+      "loss": 0.1965,
+      "step": 1082
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.48856398463249207,
+      "learning_rate": 5.0307094125298956e-05,
+      "loss": 0.2802,
+      "step": 1083
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5184898972511292,
+      "learning_rate": 5.023032122750759e-05,
+      "loss": 0.2054,
+      "step": 1084
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.49124428629875183,
+      "learning_rate": 5.015354778668805e-05,
+      "loss": 0.2585,
+      "step": 1085
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.42134425044059753,
+      "learning_rate": 5.007677398384901e-05,
+      "loss": 0.1724,
+      "step": 1086
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.4348292648792267,
+      "learning_rate": 5e-05,
+      "loss": 0.1566,
+      "step": 1087
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.39798232913017273,
+      "learning_rate": 4.9923226016151e-05,
+      "loss": 0.1677,
+      "step": 1088
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.4539756178855896,
+      "learning_rate": 4.984645221331196e-05,
+      "loss": 0.2094,
+      "step": 1089
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.43417632579803467,
+      "learning_rate": 4.976967877249242e-05,
+      "loss": 0.2205,
+      "step": 1090
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.43081367015838623,
+      "learning_rate": 4.9692905874701055e-05,
+      "loss": 0.1691,
+      "step": 1091
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.37833738327026367,
+      "learning_rate": 4.961613370094527e-05,
+      "loss": 0.1428,
+      "step": 1092
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5440829396247864,
+      "learning_rate": 4.953936243223076e-05,
+      "loss": 0.2049,
+      "step": 1093
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.42081502079963684,
+      "learning_rate": 4.9462592249561095e-05,
+      "loss": 0.1843,
+      "step": 1094
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5288732051849365,
+      "learning_rate": 4.938582333393727e-05,
+      "loss": 0.2386,
+      "step": 1095
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.4220171272754669,
+      "learning_rate": 4.9309055866357295e-05,
+      "loss": 0.1654,
+      "step": 1096
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.4985312223434448,
+      "learning_rate": 4.923229002781577e-05,
+      "loss": 0.2065,
+      "step": 1097
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.45747068524360657,
+      "learning_rate": 4.915552599930345e-05,
+      "loss": 0.1634,
+      "step": 1098
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.3894186019897461,
+      "learning_rate": 4.907876396180684e-05,
+      "loss": 0.1404,
+      "step": 1099
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.4597219228744507,
+      "learning_rate": 4.900200409630771e-05,
+      "loss": 0.1983,
+      "step": 1100
+    },
+    {
+      "epoch": 1.04,
+      "eval_loss": 0.29543277621269226,
+      "eval_runtime": 365.5253,
+      "eval_samples_per_second": 0.933,
+      "eval_steps_per_second": 0.235,
+      "step": 1100
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.4925377368927002,
+      "learning_rate": 4.892524658378276e-05,
+      "loss": 0.2211,
+      "step": 1101
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.400172621011734,
+      "learning_rate": 4.884849160520311e-05,
+      "loss": 0.1589,
+      "step": 1102
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5613206028938293,
+      "learning_rate": 4.8771739341533916e-05,
+      "loss": 0.224,
+      "step": 1103
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.44290465116500854,
+      "learning_rate": 4.869498997373393e-05,
+      "loss": 0.1805,
+      "step": 1104
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.4487089216709137,
+      "learning_rate": 4.8618243682755074e-05,
+      "loss": 0.1849,
+      "step": 1105
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5316553115844727,
+      "learning_rate": 4.8541500649542014e-05,
+      "loss": 0.1636,
+      "step": 1106
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6391955018043518,
+      "learning_rate": 4.8464761055031755e-05,
+      "loss": 0.264,
+      "step": 1107
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6015205979347229,
+      "learning_rate": 4.838802508015316e-05,
+      "loss": 0.2214,
+      "step": 1108
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.565466582775116,
+      "learning_rate": 4.831129290582659e-05,
+      "loss": 0.1871,
+      "step": 1109
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.4852063059806824,
+      "learning_rate": 4.823456471296344e-05,
+      "loss": 0.1395,
+      "step": 1110
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5589193105697632,
+      "learning_rate": 4.815784068246571e-05,
+      "loss": 0.2266,
+      "step": 1111
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5633283257484436,
+      "learning_rate": 4.8081120995225574e-05,
+      "loss": 0.23,
+      "step": 1112
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5570093393325806,
+      "learning_rate": 4.800440583212499e-05,
+      "loss": 0.2651,
+      "step": 1113
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.46370866894721985,
+      "learning_rate": 4.792769537403523e-05,
+      "loss": 0.1662,
+      "step": 1114
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5438929796218872,
+      "learning_rate": 4.785098980181649e-05,
+      "loss": 0.2378,
+      "step": 1115
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.4021894037723541,
+      "learning_rate": 4.777428929631743e-05,
+      "loss": 0.1229,
+      "step": 1116
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.46604400873184204,
+      "learning_rate": 4.7697594038374786e-05,
+      "loss": 0.1933,
+      "step": 1117
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5565809607505798,
+      "learning_rate": 4.762090420881289e-05,
+      "loss": 0.2224,
+      "step": 1118
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5792210102081299,
+      "learning_rate": 4.754421998844331e-05,
+      "loss": 0.2537,
+      "step": 1119
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.4428657591342926,
+      "learning_rate": 4.746754155806437e-05,
+      "loss": 0.1594,
+      "step": 1120
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5800593495368958,
+      "learning_rate": 4.7390869098460745e-05,
+      "loss": 0.1878,
+      "step": 1121
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.42508307099342346,
+      "learning_rate": 4.731420279040303e-05,
+      "loss": 0.1709,
+      "step": 1122
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5893201231956482,
+      "learning_rate": 4.7237542814647325e-05,
+      "loss": 0.2425,
+      "step": 1123
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5533030033111572,
+      "learning_rate": 4.716088935193479e-05,
+      "loss": 0.2271,
+      "step": 1124
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4818360209465027,
+      "learning_rate": 4.708424258299124e-05,
+      "loss": 0.1967,
+      "step": 1125
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.6182150840759277,
+      "learning_rate": 4.7007602688526695e-05,
+      "loss": 0.2371,
+      "step": 1126
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.454012006521225,
+      "learning_rate": 4.693096984923499e-05,
+      "loss": 0.1338,
+      "step": 1127
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.555012583732605,
+      "learning_rate": 4.6854344245793305e-05,
+      "loss": 0.1924,
+      "step": 1128
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.614747941493988,
+      "learning_rate": 4.677772605886175e-05,
+      "loss": 0.2312,
+      "step": 1129
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5089842081069946,
+      "learning_rate": 4.6701115469083e-05,
+      "loss": 0.1956,
+      "step": 1130
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.640500009059906,
+      "learning_rate": 4.662451265708174e-05,
+      "loss": 0.2569,
+      "step": 1131
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.6651586294174194,
+      "learning_rate": 4.65479178034644e-05,
+      "loss": 0.2226,
+      "step": 1132
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5069378018379211,
+      "learning_rate": 4.647133108881858e-05,
+      "loss": 0.1936,
+      "step": 1133
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5977818965911865,
+      "learning_rate": 4.6394752693712734e-05,
+      "loss": 0.2329,
+      "step": 1134
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5742204189300537,
+      "learning_rate": 4.631818279869566e-05,
+      "loss": 0.2213,
+      "step": 1135
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6270827054977417,
+      "learning_rate": 4.624162158429619e-05,
+      "loss": 0.262,
+      "step": 1136
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5989996790885925,
+      "learning_rate": 4.616506923102259e-05,
+      "loss": 0.1967,
+      "step": 1137
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.4012334942817688,
+      "learning_rate": 4.608852591936231e-05,
+      "loss": 0.1319,
+      "step": 1138
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6948206424713135,
+      "learning_rate": 4.6011991829781456e-05,
+      "loss": 0.2809,
+      "step": 1139
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5710472464561462,
+      "learning_rate": 4.593546714272438e-05,
+      "loss": 0.2278,
+      "step": 1140
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6348649859428406,
+      "learning_rate": 4.585895203861328e-05,
+      "loss": 0.2224,
+      "step": 1141
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6497195959091187,
+      "learning_rate": 4.578244669784777e-05,
+      "loss": 0.2442,
+      "step": 1142
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6004654765129089,
+      "learning_rate": 4.5705951300804424e-05,
+      "loss": 0.2969,
+      "step": 1143
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5684016942977905,
+      "learning_rate": 4.562946602783636e-05,
+      "loss": 0.1767,
+      "step": 1144
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.7365360260009766,
+      "learning_rate": 4.55529910592729e-05,
+      "loss": 0.221,
+      "step": 1145
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5203796029090881,
+      "learning_rate": 4.547652657541897e-05,
+      "loss": 0.1509,
+      "step": 1146
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.4253897964954376,
+      "learning_rate": 4.540007275655485e-05,
+      "loss": 0.1651,
+      "step": 1147
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5274244546890259,
+      "learning_rate": 4.532362978293564e-05,
+      "loss": 0.2418,
+      "step": 1148
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.4286225736141205,
+      "learning_rate": 4.5247197834790876e-05,
+      "loss": 0.1704,
+      "step": 1149
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.42857682704925537,
+      "learning_rate": 4.517077709232411e-05,
+      "loss": 0.1611,
+      "step": 1150
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.4747050106525421,
+      "learning_rate": 4.509436773571247e-05,
+      "loss": 0.2065,
+      "step": 1151
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.4124803841114044,
+      "learning_rate": 4.501796994510622e-05,
+      "loss": 0.1561,
+      "step": 1152
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5347474217414856,
+      "learning_rate": 4.49415839006284e-05,
+      "loss": 0.2062,
+      "step": 1153
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.511516809463501,
+      "learning_rate": 4.4865209782374304e-05,
+      "loss": 0.194,
+      "step": 1154
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.46989378333091736,
+      "learning_rate": 4.478884777041115e-05,
+      "loss": 0.1956,
+      "step": 1155
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5284501910209656,
+      "learning_rate": 4.471249804477758e-05,
+      "loss": 0.1885,
+      "step": 1156
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.506072998046875,
+      "learning_rate": 4.463616078548328e-05,
+      "loss": 0.2557,
+      "step": 1157
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4575859010219574,
+      "learning_rate": 4.4559836172508565e-05,
+      "loss": 0.1861,
+      "step": 1158
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5177057385444641,
+      "learning_rate": 4.448352438580391e-05,
+      "loss": 0.2737,
+      "step": 1159
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4681926965713501,
+      "learning_rate": 4.4407225605289547e-05,
+      "loss": 0.1987,
+      "step": 1160
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5116195678710938,
+      "learning_rate": 4.433094001085505e-05,
+      "loss": 0.2086,
+      "step": 1161
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.46723708510398865,
+      "learning_rate": 4.4254667782358924e-05,
+      "loss": 0.1769,
+      "step": 1162
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.49821409583091736,
+      "learning_rate": 4.417840909962813e-05,
+      "loss": 0.2075,
+      "step": 1163
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5224546194076538,
+      "learning_rate": 4.410216414245771e-05,
+      "loss": 0.2546,
+      "step": 1164
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.4872426688671112,
+      "learning_rate": 4.402593309061034e-05,
+      "loss": 0.1727,
+      "step": 1165
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.7369954586029053,
+      "learning_rate": 4.394971612381591e-05,
+      "loss": 0.2346,
+      "step": 1166
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.4489041864871979,
+      "learning_rate": 4.38735134217711e-05,
+      "loss": 0.1352,
+      "step": 1167
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.49813026189804077,
+      "learning_rate": 4.379732516413897e-05,
+      "loss": 0.1736,
+      "step": 1168
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.43354296684265137,
+      "learning_rate": 4.37211515305485e-05,
+      "loss": 0.1307,
+      "step": 1169
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5650145411491394,
+      "learning_rate": 4.364499270059424e-05,
+      "loss": 0.1875,
+      "step": 1170
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5466250777244568,
+      "learning_rate": 4.3568848853835774e-05,
+      "loss": 0.2028,
+      "step": 1171
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.44607916474342346,
+      "learning_rate": 4.34927201697974e-05,
+      "loss": 0.1618,
+      "step": 1172
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5923343896865845,
+      "learning_rate": 4.341660682796766e-05,
+      "loss": 0.1733,
+      "step": 1173
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5762339234352112,
+      "learning_rate": 4.334050900779893e-05,
+      "loss": 0.2109,
+      "step": 1174
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5711573958396912,
+      "learning_rate": 4.3264426888706965e-05,
+      "loss": 0.2243,
+      "step": 1175
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5193672776222229,
+      "learning_rate": 4.3188360650070525e-05,
+      "loss": 0.1957,
+      "step": 1176
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5383495092391968,
+      "learning_rate": 4.311231047123092e-05,
+      "loss": 0.1878,
+      "step": 1177
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.48797503113746643,
+      "learning_rate": 4.303627653149159e-05,
+      "loss": 0.1696,
+      "step": 1178
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5811154246330261,
+      "learning_rate": 4.2960259010117724e-05,
+      "loss": 0.1782,
+      "step": 1179
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.44567590951919556,
+      "learning_rate": 4.288425808633575e-05,
+      "loss": 0.1488,
+      "step": 1180
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.4402742087841034,
+      "learning_rate": 4.2808273939333e-05,
+      "loss": 0.1177,
+      "step": 1181
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.575922429561615,
+      "learning_rate": 4.2732306748257224e-05,
+      "loss": 0.1875,
+      "step": 1182
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5445305109024048,
+      "learning_rate": 4.265635669221622e-05,
+      "loss": 0.2101,
+      "step": 1183
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5818389058113098,
+      "learning_rate": 4.258042395027738e-05,
+      "loss": 0.1747,
+      "step": 1184
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5533482432365417,
+      "learning_rate": 4.250450870146725e-05,
+      "loss": 0.203,
+      "step": 1185
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.4254933297634125,
+      "learning_rate": 4.2428611124771184e-05,
+      "loss": 0.1813,
+      "step": 1186
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5498292446136475,
+      "learning_rate": 4.2352731399132804e-05,
+      "loss": 0.1925,
+      "step": 1187
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.6894237995147705,
+      "learning_rate": 4.2276869703453724e-05,
+      "loss": 0.2316,
+      "step": 1188
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5249069333076477,
+      "learning_rate": 4.2201026216592975e-05,
+      "loss": 0.2258,
+      "step": 1189
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5850760340690613,
+      "learning_rate": 4.21252011173667e-05,
+      "loss": 0.268,
+      "step": 1190
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.42634671926498413,
+      "learning_rate": 4.2049394584547664e-05,
+      "loss": 0.1115,
+      "step": 1191
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.52420973777771,
+      "learning_rate": 4.1973606796864884e-05,
+      "loss": 0.1777,
+      "step": 1192
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5667282938957214,
+      "learning_rate": 4.189783793300316e-05,
+      "loss": 0.2373,
+      "step": 1193
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.47565990686416626,
+      "learning_rate": 4.182208817160269e-05,
+      "loss": 0.1349,
+      "step": 1194
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.6549266576766968,
+      "learning_rate": 4.1746357691258614e-05,
+      "loss": 0.1769,
+      "step": 1195
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5758785009384155,
+      "learning_rate": 4.167064667052065e-05,
+      "loss": 0.2281,
+      "step": 1196
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.501886248588562,
+      "learning_rate": 4.15949552878926e-05,
+      "loss": 0.1331,
+      "step": 1197
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5422622561454773,
+      "learning_rate": 4.151928372183198e-05,
+      "loss": 0.2001,
+      "step": 1198
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5602012872695923,
+      "learning_rate": 4.144363215074958e-05,
+      "loss": 0.1661,
+      "step": 1199
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5743290185928345,
+      "learning_rate": 4.136800075300906e-05,
+      "loss": 0.2242,
+      "step": 1200
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 0.30257686972618103,
+      "eval_runtime": 366.9881,
+      "eval_samples_per_second": 0.929,
+      "eval_steps_per_second": 0.234,
+      "step": 1200
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5308936834335327,
+      "learning_rate": 4.12923897069265e-05,
+      "loss": 0.1959,
+      "step": 1201
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.4427596628665924,
+      "learning_rate": 4.1216799190770006e-05,
+      "loss": 0.1462,
+      "step": 1202
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5490732789039612,
+      "learning_rate": 4.1141229382759286e-05,
+      "loss": 0.1758,
+      "step": 1203
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.47641435265541077,
+      "learning_rate": 4.1065680461065194e-05,
+      "loss": 0.1786,
+      "step": 1204
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.590862512588501,
+      "learning_rate": 4.099015260380939e-05,
+      "loss": 0.2659,
+      "step": 1205
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5282086730003357,
+      "learning_rate": 4.0914645989063845e-05,
+      "loss": 0.1872,
+      "step": 1206
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.6061357259750366,
+      "learning_rate": 4.0839160794850436e-05,
+      "loss": 0.2572,
+      "step": 1207
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.7573590278625488,
+      "learning_rate": 4.076369719914055e-05,
+      "loss": 0.3385,
+      "step": 1208
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.47287607192993164,
+      "learning_rate": 4.068825537985465e-05,
+      "loss": 0.163,
+      "step": 1209
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.47131070494651794,
+      "learning_rate": 4.061283551486185e-05,
+      "loss": 0.2229,
+      "step": 1210
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.46008923649787903,
+      "learning_rate": 4.0537437781979506e-05,
+      "loss": 0.1725,
+      "step": 1211
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.4891980290412903,
+      "learning_rate": 4.0462062358972796e-05,
+      "loss": 0.2251,
+      "step": 1212
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.4593801200389862,
+      "learning_rate": 4.038670942355431e-05,
+      "loss": 0.1609,
+      "step": 1213
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.535732090473175,
+      "learning_rate": 4.03113791533836e-05,
+      "loss": 0.2052,
+      "step": 1214
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.45214179158210754,
+      "learning_rate": 4.023607172606676e-05,
+      "loss": 0.1632,
+      "step": 1215
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.47409456968307495,
+      "learning_rate": 4.016078731915608e-05,
+      "loss": 0.1952,
+      "step": 1216
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.4977971613407135,
+      "learning_rate": 4.008552611014955e-05,
+      "loss": 0.1995,
+      "step": 1217
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.4798290729522705,
+      "learning_rate": 4.001028827649046e-05,
+      "loss": 0.2118,
+      "step": 1218
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.446363240480423,
+      "learning_rate": 3.993507399556699e-05,
+      "loss": 0.1758,
+      "step": 1219
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.46178558468818665,
+      "learning_rate": 3.9859883444711794e-05,
+      "loss": 0.1723,
+      "step": 1220
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.43594110012054443,
+      "learning_rate": 3.978471680120157e-05,
+      "loss": 0.1939,
+      "step": 1221
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5007155537605286,
+      "learning_rate": 3.970957424225666e-05,
+      "loss": 0.1947,
+      "step": 1222
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.4080187976360321,
+      "learning_rate": 3.9634455945040614e-05,
+      "loss": 0.1366,
+      "step": 1223
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.46378999948501587,
+      "learning_rate": 3.955936208665979e-05,
+      "loss": 0.1603,
+      "step": 1224
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.4351801574230194,
+      "learning_rate": 3.9484292844162904e-05,
+      "loss": 0.1784,
+      "step": 1225
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5361440777778625,
+      "learning_rate": 3.940924839454066e-05,
+      "loss": 0.2108,
+      "step": 1226
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5310454368591309,
+      "learning_rate": 3.933422891472532e-05,
+      "loss": 0.2236,
+      "step": 1227
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.47151508927345276,
+      "learning_rate": 3.925923458159023e-05,
+      "loss": 0.1596,
+      "step": 1228
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5100865364074707,
+      "learning_rate": 3.918426557194947e-05,
+      "loss": 0.1768,
+      "step": 1229
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5460057258605957,
+      "learning_rate": 3.9109322062557424e-05,
+      "loss": 0.2074,
+      "step": 1230
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.45145949721336365,
+      "learning_rate": 3.903440423010834e-05,
+      "loss": 0.1431,
+      "step": 1231
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6152063012123108,
+      "learning_rate": 3.895951225123595e-05,
+      "loss": 0.238,
+      "step": 1232
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6215837597846985,
+      "learning_rate": 3.888464630251298e-05,
+      "loss": 0.2597,
+      "step": 1233
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6012809872627258,
+      "learning_rate": 3.880980656045087e-05,
+      "loss": 0.2409,
+      "step": 1234
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.46502628922462463,
+      "learning_rate": 3.873499320149918e-05,
+      "loss": 0.1372,
+      "step": 1235
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.4605456292629242,
+      "learning_rate": 3.866020640204533e-05,
+      "loss": 0.1592,
+      "step": 1236
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6536608338356018,
+      "learning_rate": 3.858544633841409e-05,
+      "loss": 0.2739,
+      "step": 1237
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6082574129104614,
+      "learning_rate": 3.851071318686721e-05,
+      "loss": 0.1744,
+      "step": 1238
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6058924198150635,
+      "learning_rate": 3.843600712360298e-05,
+      "loss": 0.1515,
+      "step": 1239
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.4204099476337433,
+      "learning_rate": 3.8361328324755825e-05,
+      "loss": 0.1208,
+      "step": 1240
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.7002831697463989,
+      "learning_rate": 3.82866769663959e-05,
+      "loss": 0.2715,
+      "step": 1241
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5566287636756897,
+      "learning_rate": 3.821205322452863e-05,
+      "loss": 0.2108,
+      "step": 1242
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6842728853225708,
+      "learning_rate": 3.813745727509439e-05,
+      "loss": 0.2911,
+      "step": 1243
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.4805491864681244,
+      "learning_rate": 3.8062889293967976e-05,
+      "loss": 0.205,
+      "step": 1244
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6155717968940735,
+      "learning_rate": 3.798834945695826e-05,
+      "loss": 0.2178,
+      "step": 1245
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.53364497423172,
+      "learning_rate": 3.791383793980776e-05,
+      "loss": 0.2357,
+      "step": 1246
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6079195141792297,
+      "learning_rate": 3.7839354918192215e-05,
+      "loss": 0.2571,
+      "step": 1247
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3689627945423126,
+      "learning_rate": 3.7764900567720196e-05,
+      "loss": 0.1274,
+      "step": 1248
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5014574527740479,
+      "learning_rate": 3.769047506393267e-05,
+      "loss": 0.1805,
+      "step": 1249
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.45645275712013245,
+      "learning_rate": 3.761607858230257e-05,
+      "loss": 0.1507,
+      "step": 1250
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5433384776115417,
+      "learning_rate": 3.7541711298234436e-05,
+      "loss": 0.2515,
+      "step": 1251
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.44526979327201843,
+      "learning_rate": 3.746737338706397e-05,
+      "loss": 0.1462,
+      "step": 1252
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5026818513870239,
+      "learning_rate": 3.7393065024057596e-05,
+      "loss": 0.1503,
+      "step": 1253
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6757113933563232,
+      "learning_rate": 3.7318786384412074e-05,
+      "loss": 0.2821,
+      "step": 1254
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.43436071276664734,
+      "learning_rate": 3.724453764325411e-05,
+      "loss": 0.1726,
+      "step": 1255
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.4763764441013336,
+      "learning_rate": 3.71703189756399e-05,
+      "loss": 0.1763,
+      "step": 1256
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5709313750267029,
+      "learning_rate": 3.709613055655474e-05,
+      "loss": 0.2271,
+      "step": 1257
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.48193347454071045,
+      "learning_rate": 3.70219725609126e-05,
+      "loss": 0.2024,
+      "step": 1258
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.47502586245536804,
+      "learning_rate": 3.694784516355573e-05,
+      "loss": 0.1775,
+      "step": 1259
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5688880085945129,
+      "learning_rate": 3.6873748539254246e-05,
+      "loss": 0.1677,
+      "step": 1260
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.44192633032798767,
+      "learning_rate": 3.679968286270571e-05,
+      "loss": 0.1565,
+      "step": 1261
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.43847915530204773,
+      "learning_rate": 3.6725648308534696e-05,
+      "loss": 0.1545,
+      "step": 1262
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.575657069683075,
+      "learning_rate": 3.6651645051292415e-05,
+      "loss": 0.1717,
+      "step": 1263
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6538360714912415,
+      "learning_rate": 3.65776732654563e-05,
+      "loss": 0.2788,
+      "step": 1264
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6082655191421509,
+      "learning_rate": 3.650373312542956e-05,
+      "loss": 0.2236,
+      "step": 1265
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5325061678886414,
+      "learning_rate": 3.6429824805540815e-05,
+      "loss": 0.2072,
+      "step": 1266
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5524687170982361,
+      "learning_rate": 3.6355948480043644e-05,
+      "loss": 0.2185,
+      "step": 1267
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6572748422622681,
+      "learning_rate": 3.628210432311621e-05,
+      "loss": 0.3272,
+      "step": 1268
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5233578085899353,
+      "learning_rate": 3.620829250886083e-05,
+      "loss": 0.2055,
+      "step": 1269
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.4601766765117645,
+      "learning_rate": 3.6134513211303556e-05,
+      "loss": 0.1664,
+      "step": 1270
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.4763219952583313,
+      "learning_rate": 3.606076660439378e-05,
+      "loss": 0.1645,
+      "step": 1271
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5625105500221252,
+      "learning_rate": 3.598705286200382e-05,
+      "loss": 0.2475,
+      "step": 1272
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5452956557273865,
+      "learning_rate": 3.591337215792852e-05,
+      "loss": 0.1691,
+      "step": 1273
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5354436039924622,
+      "learning_rate": 3.58397246658848e-05,
+      "loss": 0.2009,
+      "step": 1274
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5049367547035217,
+      "learning_rate": 3.576611055951131e-05,
+      "loss": 0.1791,
+      "step": 1275
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.6478241086006165,
+      "learning_rate": 3.5692530012367955e-05,
+      "loss": 0.234,
+      "step": 1276
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.4778873026371002,
+      "learning_rate": 3.561898319793555e-05,
+      "loss": 0.1799,
+      "step": 1277
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.49865901470184326,
+      "learning_rate": 3.554547028961537e-05,
+      "loss": 0.2201,
+      "step": 1278
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5965945720672607,
+      "learning_rate": 3.547199146072873e-05,
+      "loss": 0.1841,
+      "step": 1279
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.4726731777191162,
+      "learning_rate": 3.5398546884516604e-05,
+      "loss": 0.1691,
+      "step": 1280
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5452727675437927,
+      "learning_rate": 3.5325136734139214e-05,
+      "loss": 0.199,
+      "step": 1281
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5570040345191956,
+      "learning_rate": 3.5251761182675625e-05,
+      "loss": 0.2169,
+      "step": 1282
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5860921740531921,
+      "learning_rate": 3.5178420403123305e-05,
+      "loss": 0.1685,
+      "step": 1283
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5538284182548523,
+      "learning_rate": 3.510511456839777e-05,
+      "loss": 0.2232,
+      "step": 1284
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5518790483474731,
+      "learning_rate": 3.5031843851332104e-05,
+      "loss": 0.209,
+      "step": 1285
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.438111811876297,
+      "learning_rate": 3.495860842467664e-05,
+      "loss": 0.1356,
+      "step": 1286
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5282179117202759,
+      "learning_rate": 3.488540846109849e-05,
+      "loss": 0.2428,
+      "step": 1287
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.3734651803970337,
+      "learning_rate": 3.481224413318114e-05,
+      "loss": 0.144,
+      "step": 1288
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.46726593375205994,
+      "learning_rate": 3.473911561342408e-05,
+      "loss": 0.1694,
+      "step": 1289
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5814958214759827,
+      "learning_rate": 3.4666023074242356e-05,
+      "loss": 0.1849,
+      "step": 1290
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5666380524635315,
+      "learning_rate": 3.459296668796619e-05,
+      "loss": 0.256,
+      "step": 1291
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.6503203511238098,
+      "learning_rate": 3.4519946626840573e-05,
+      "loss": 0.255,
+      "step": 1292
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5842156410217285,
+      "learning_rate": 3.444696306302485e-05,
+      "loss": 0.1797,
+      "step": 1293
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.6458174586296082,
+      "learning_rate": 3.4374016168592296e-05,
+      "loss": 0.2615,
+      "step": 1294
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5390830039978027,
+      "learning_rate": 3.4301106115529764e-05,
+      "loss": 0.2177,
+      "step": 1295
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.40310320258140564,
+      "learning_rate": 3.422823307573722e-05,
+      "loss": 0.1174,
+      "step": 1296
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.4604451358318329,
+      "learning_rate": 3.41553972210274e-05,
+      "loss": 0.1751,
+      "step": 1297
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.532292366027832,
+      "learning_rate": 3.40825987231253e-05,
+      "loss": 0.1809,
+      "step": 1298
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.6451547145843506,
+      "learning_rate": 3.4009837753667916e-05,
+      "loss": 0.2221,
+      "step": 1299
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.4820696711540222,
+      "learning_rate": 3.393711448420372e-05,
+      "loss": 0.1672,
+      "step": 1300
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.30080530047416687,
+      "eval_runtime": 361.3397,
+      "eval_samples_per_second": 0.944,
+      "eval_steps_per_second": 0.238,
+      "step": 1300
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5464685559272766,
+      "learning_rate": 3.38644290861923e-05,
+      "loss": 0.2098,
+      "step": 1301
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.6084319949150085,
+      "learning_rate": 3.379178173100396e-05,
+      "loss": 0.232,
+      "step": 1302
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5863743424415588,
+      "learning_rate": 3.3719172589919326e-05,
+      "loss": 0.2188,
+      "step": 1303
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6455075144767761,
+      "learning_rate": 3.364660183412892e-05,
+      "loss": 0.2289,
+      "step": 1304
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6269391179084778,
+      "learning_rate": 3.357406963473274e-05,
+      "loss": 0.2378,
+      "step": 1305
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6516079902648926,
+      "learning_rate": 3.3501576162739904e-05,
+      "loss": 0.2392,
+      "step": 1306
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6414124369621277,
+      "learning_rate": 3.3429121589068215e-05,
+      "loss": 0.2388,
+      "step": 1307
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6592124700546265,
+      "learning_rate": 3.335670608454376e-05,
+      "loss": 0.2542,
+      "step": 1308
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.49166423082351685,
+      "learning_rate": 3.3284329819900525e-05,
+      "loss": 0.1591,
+      "step": 1309
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5908455848693848,
+      "learning_rate": 3.321199296577998e-05,
+      "loss": 0.203,
+      "step": 1310
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5878795981407166,
+      "learning_rate": 3.3139695692730646e-05,
+      "loss": 0.2257,
+      "step": 1311
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6520990133285522,
+      "learning_rate": 3.3067438171207766e-05,
+      "loss": 0.2378,
+      "step": 1312
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5301094055175781,
+      "learning_rate": 3.2995220571572846e-05,
+      "loss": 0.1912,
+      "step": 1313
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4721067547798157,
+      "learning_rate": 3.292304306409325e-05,
+      "loss": 0.1932,
+      "step": 1314
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5222894549369812,
+      "learning_rate": 3.285090581894185e-05,
+      "loss": 0.1873,
+      "step": 1315
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5250543355941772,
+      "learning_rate": 3.2778809006196564e-05,
+      "loss": 0.202,
+      "step": 1316
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.44919174909591675,
+      "learning_rate": 3.270675279583998e-05,
+      "loss": 0.1633,
+      "step": 1317
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5511341691017151,
+      "learning_rate": 3.263473735775899e-05,
+      "loss": 0.1966,
+      "step": 1318
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.6023682355880737,
+      "learning_rate": 3.2562762861744326e-05,
+      "loss": 0.2713,
+      "step": 1319
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5638202428817749,
+      "learning_rate": 3.249082947749019e-05,
+      "loss": 0.2558,
+      "step": 1320
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.6085641980171204,
+      "learning_rate": 3.2418937374593894e-05,
+      "loss": 0.2201,
+      "step": 1321
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5487499237060547,
+      "learning_rate": 3.234708672255538e-05,
+      "loss": 0.2098,
+      "step": 1322
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3549087643623352,
+      "learning_rate": 3.227527769077687e-05,
+      "loss": 0.1253,
+      "step": 1323
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5621610283851624,
+      "learning_rate": 3.220351044856247e-05,
+      "loss": 0.1825,
+      "step": 1324
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5300235748291016,
+      "learning_rate": 3.2131785165117745e-05,
+      "loss": 0.1747,
+      "step": 1325
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.6438629031181335,
+      "learning_rate": 3.206010200954935e-05,
+      "loss": 0.1862,
+      "step": 1326
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5685358643531799,
+      "learning_rate": 3.198846115086459e-05,
+      "loss": 0.2314,
+      "step": 1327
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.602911114692688,
+      "learning_rate": 3.191686275797107e-05,
+      "loss": 0.1826,
+      "step": 1328
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.528223991394043,
+      "learning_rate": 3.1845306999676275e-05,
+      "loss": 0.205,
+      "step": 1329
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5360718965530396,
+      "learning_rate": 3.177379404468715e-05,
+      "loss": 0.1892,
+      "step": 1330
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.6214163899421692,
+      "learning_rate": 3.1702324061609736e-05,
+      "loss": 0.2313,
+      "step": 1331
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.47413507103919983,
+      "learning_rate": 3.163089721894876e-05,
+      "loss": 0.1208,
+      "step": 1332
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5803177952766418,
+      "learning_rate": 3.1559513685107235e-05,
+      "loss": 0.2237,
+      "step": 1333
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.40417158603668213,
+      "learning_rate": 3.148817362838606e-05,
+      "loss": 0.127,
+      "step": 1334
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.63303542137146,
+      "learning_rate": 3.141687721698363e-05,
+      "loss": 0.1582,
+      "step": 1335
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.525445282459259,
+      "learning_rate": 3.134562461899545e-05,
+      "loss": 0.198,
+      "step": 1336
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5178670287132263,
+      "learning_rate": 3.127441600241369e-05,
+      "loss": 0.1816,
+      "step": 1337
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5207834243774414,
+      "learning_rate": 3.120325153512687e-05,
+      "loss": 0.2045,
+      "step": 1338
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.4817272424697876,
+      "learning_rate": 3.11321313849194e-05,
+      "loss": 0.164,
+      "step": 1339
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.668614387512207,
+      "learning_rate": 3.10610557194712e-05,
+      "loss": 0.2379,
+      "step": 1340
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.47183382511138916,
+      "learning_rate": 3.0990024706357314e-05,
+      "loss": 0.1675,
+      "step": 1341
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.617523193359375,
+      "learning_rate": 3.091903851304751e-05,
+      "loss": 0.2005,
+      "step": 1342
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5135393142700195,
+      "learning_rate": 3.084809730690587e-05,
+      "loss": 0.1751,
+      "step": 1343
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5151330232620239,
+      "learning_rate": 3.077720125519042e-05,
+      "loss": 0.1795,
+      "step": 1344
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.6077806949615479,
+      "learning_rate": 3.070635052505273e-05,
+      "loss": 0.2254,
+      "step": 1345
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.49431484937667847,
+      "learning_rate": 3.063554528353752e-05,
+      "loss": 0.1758,
+      "step": 1346
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.4728491008281708,
+      "learning_rate": 3.056478569758225e-05,
+      "loss": 0.1682,
+      "step": 1347
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5673767328262329,
+      "learning_rate": 3.0494071934016737e-05,
+      "loss": 0.1827,
+      "step": 1348
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5223566889762878,
+      "learning_rate": 3.0423404159562774e-05,
+      "loss": 0.1777,
+      "step": 1349
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.6421141624450684,
+      "learning_rate": 3.03527825408337e-05,
+      "loss": 0.1863,
+      "step": 1350
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5192697048187256,
+      "learning_rate": 3.0282207244334082e-05,
+      "loss": 0.1962,
+      "step": 1351
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5397422909736633,
+      "learning_rate": 3.0211678436459212e-05,
+      "loss": 0.2026,
+      "step": 1352
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5058582425117493,
+      "learning_rate": 3.014119628349482e-05,
+      "loss": 0.1769,
+      "step": 1353
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.441275417804718,
+      "learning_rate": 3.0070760951616618e-05,
+      "loss": 0.1602,
+      "step": 1354
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5515382289886475,
+      "learning_rate": 3.0000372606889937e-05,
+      "loss": 0.1877,
+      "step": 1355
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.4784175455570221,
+      "learning_rate": 2.993003141526933e-05,
+      "loss": 0.1411,
+      "step": 1356
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.43732354044914246,
+      "learning_rate": 2.9859737542598155e-05,
+      "loss": 0.1497,
+      "step": 1357
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.4430086612701416,
+      "learning_rate": 2.978949115460824e-05,
+      "loss": 0.1526,
+      "step": 1358
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5904696583747864,
+      "learning_rate": 2.971929241691942e-05,
+      "loss": 0.2383,
+      "step": 1359
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.46987247467041016,
+      "learning_rate": 2.9649141495039223e-05,
+      "loss": 0.2018,
+      "step": 1360
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.4308496415615082,
+      "learning_rate": 2.957903855436241e-05,
+      "loss": 0.0978,
+      "step": 1361
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6394721269607544,
+      "learning_rate": 2.950898376017064e-05,
+      "loss": 0.2999,
+      "step": 1362
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.38467690348625183,
+      "learning_rate": 2.9438977277632017e-05,
+      "loss": 0.1029,
+      "step": 1363
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6654496192932129,
+      "learning_rate": 2.9369019271800823e-05,
+      "loss": 0.2954,
+      "step": 1364
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5223631262779236,
+      "learning_rate": 2.9299109907616955e-05,
+      "loss": 0.1711,
+      "step": 1365
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5331325531005859,
+      "learning_rate": 2.9229249349905684e-05,
+      "loss": 0.1793,
+      "step": 1366
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.6004540920257568,
+      "learning_rate": 2.9159437763377184e-05,
+      "loss": 0.2423,
+      "step": 1367
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.45279252529144287,
+      "learning_rate": 2.908967531262618e-05,
+      "loss": 0.1232,
+      "step": 1368
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4995564818382263,
+      "learning_rate": 2.9019962162131565e-05,
+      "loss": 0.175,
+      "step": 1369
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4184959828853607,
+      "learning_rate": 2.895029847625595e-05,
+      "loss": 0.138,
+      "step": 1370
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4497951865196228,
+      "learning_rate": 2.8880684419245384e-05,
+      "loss": 0.1239,
+      "step": 1371
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.47717565298080444,
+      "learning_rate": 2.8811120155228844e-05,
+      "loss": 0.1735,
+      "step": 1372
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4506407678127289,
+      "learning_rate": 2.874160584821798e-05,
+      "loss": 0.1633,
+      "step": 1373
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.48292016983032227,
+      "learning_rate": 2.8672141662106578e-05,
+      "loss": 0.1346,
+      "step": 1374
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.597580075263977,
+      "learning_rate": 2.8602727760670332e-05,
+      "loss": 0.2393,
+      "step": 1375
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4248458743095398,
+      "learning_rate": 2.853336430756631e-05,
+      "loss": 0.1383,
+      "step": 1376
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.47254353761672974,
+      "learning_rate": 2.846405146633269e-05,
+      "loss": 0.1584,
+      "step": 1377
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5984407067298889,
+      "learning_rate": 2.8394789400388328e-05,
+      "loss": 0.2177,
+      "step": 1378
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5190450549125671,
+      "learning_rate": 2.8325578273032294e-05,
+      "loss": 0.1585,
+      "step": 1379
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.6197163462638855,
+      "learning_rate": 2.8256418247443662e-05,
+      "loss": 0.1993,
+      "step": 1380
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.6294987201690674,
+      "learning_rate": 2.8187309486680925e-05,
+      "loss": 0.2154,
+      "step": 1381
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5441323518753052,
+      "learning_rate": 2.811825215368179e-05,
+      "loss": 0.1679,
+      "step": 1382
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5318465232849121,
+      "learning_rate": 2.8049246411262643e-05,
+      "loss": 0.1558,
+      "step": 1383
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5266247987747192,
+      "learning_rate": 2.798029242211828e-05,
+      "loss": 0.1594,
+      "step": 1384
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5665544867515564,
+      "learning_rate": 2.7911390348821508e-05,
+      "loss": 0.1676,
+      "step": 1385
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5930096507072449,
+      "learning_rate": 2.7842540353822632e-05,
+      "loss": 0.2034,
+      "step": 1386
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.6868606805801392,
+      "learning_rate": 2.777374259944929e-05,
+      "loss": 0.2541,
+      "step": 1387
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.4904882609844208,
+      "learning_rate": 2.770499724790584e-05,
+      "loss": 0.1466,
+      "step": 1388
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6401095986366272,
+      "learning_rate": 2.7636304461273187e-05,
+      "loss": 0.1825,
+      "step": 1389
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5682973861694336,
+      "learning_rate": 2.7567664401508225e-05,
+      "loss": 0.1505,
+      "step": 1390
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5290000438690186,
+      "learning_rate": 2.7499077230443603e-05,
+      "loss": 0.1587,
+      "step": 1391
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5009942650794983,
+      "learning_rate": 2.7430543109787222e-05,
+      "loss": 0.1382,
+      "step": 1392
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6826687455177307,
+      "learning_rate": 2.7362062201121918e-05,
+      "loss": 0.2104,
+      "step": 1393
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5887047648429871,
+      "learning_rate": 2.729363466590511e-05,
+      "loss": 0.1579,
+      "step": 1394
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.48021191358566284,
+      "learning_rate": 2.72252606654683e-05,
+      "loss": 0.1378,
+      "step": 1395
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6405319571495056,
+      "learning_rate": 2.7156940361016864e-05,
+      "loss": 0.2173,
+      "step": 1396
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6246445775032043,
+      "learning_rate": 2.708867391362948e-05,
+      "loss": 0.2015,
+      "step": 1397
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6238641142845154,
+      "learning_rate": 2.702046148425795e-05,
+      "loss": 0.2298,
+      "step": 1398
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.4616875946521759,
+      "learning_rate": 2.6952303233726627e-05,
+      "loss": 0.1095,
+      "step": 1399
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5137340426445007,
+      "learning_rate": 2.688419932273219e-05,
+      "loss": 0.1898,
+      "step": 1400
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 0.3004399538040161,
+      "eval_runtime": 366.7299,
+      "eval_samples_per_second": 0.93,
+      "eval_steps_per_second": 0.235,
+      "step": 1400
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5037074685096741,
+      "learning_rate": 2.6816149911843148e-05,
+      "loss": 0.1389,
+      "step": 1401
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.49550512433052063,
+      "learning_rate": 2.6748155161499567e-05,
+      "loss": 0.1431,
+      "step": 1402
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6803264617919922,
+      "learning_rate": 2.6680215232012628e-05,
+      "loss": 0.2479,
+      "step": 1403
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5337669253349304,
+      "learning_rate": 2.6612330283564223e-05,
+      "loss": 0.1933,
+      "step": 1404
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5505081415176392,
+      "learning_rate": 2.6544500476206673e-05,
+      "loss": 0.2031,
+      "step": 1405
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.41485393047332764,
+      "learning_rate": 2.6476725969862227e-05,
+      "loss": 0.1119,
+      "step": 1406
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5651363730430603,
+      "learning_rate": 2.640900692432282e-05,
+      "loss": 0.2032,
+      "step": 1407
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5666103959083557,
+      "learning_rate": 2.634134349924956e-05,
+      "loss": 0.1774,
+      "step": 1408
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6503807306289673,
+      "learning_rate": 2.6273735854172488e-05,
+      "loss": 0.2045,
+      "step": 1409
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.49487918615341187,
+      "learning_rate": 2.6206184148490066e-05,
+      "loss": 0.19,
+      "step": 1410
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6030619144439697,
+      "learning_rate": 2.6138688541468903e-05,
+      "loss": 0.2429,
+      "step": 1411
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.4683573544025421,
+      "learning_rate": 2.6071249192243363e-05,
+      "loss": 0.1432,
+      "step": 1412
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6380842328071594,
+      "learning_rate": 2.6003866259815123e-05,
+      "loss": 0.2337,
+      "step": 1413
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5573086142539978,
+      "learning_rate": 2.5936539903052892e-05,
+      "loss": 0.2471,
+      "step": 1414
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5931753516197205,
+      "learning_rate": 2.5869270280691944e-05,
+      "loss": 0.1715,
+      "step": 1415
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6511543393135071,
+      "learning_rate": 2.580205755133384e-05,
+      "loss": 0.2027,
+      "step": 1416
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.4882636070251465,
+      "learning_rate": 2.573490187344596e-05,
+      "loss": 0.1321,
+      "step": 1417
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.39837753772735596,
+      "learning_rate": 2.5667803405361213e-05,
+      "loss": 0.1384,
+      "step": 1418
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.48738205432891846,
+      "learning_rate": 2.5600762305277583e-05,
+      "loss": 0.1835,
+      "step": 1419
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5051771402359009,
+      "learning_rate": 2.5533778731257824e-05,
+      "loss": 0.1107,
+      "step": 1420
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5008789896965027,
+      "learning_rate": 2.5466852841229088e-05,
+      "loss": 0.1426,
+      "step": 1421
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5426428318023682,
+      "learning_rate": 2.5399984792982457e-05,
+      "loss": 0.2107,
+      "step": 1422
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.48200392723083496,
+      "learning_rate": 2.5333174744172704e-05,
+      "loss": 0.1613,
+      "step": 1423
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.4660708010196686,
+      "learning_rate": 2.5266422852317795e-05,
+      "loss": 0.1609,
+      "step": 1424
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6523826122283936,
+      "learning_rate": 2.5199729274798662e-05,
+      "loss": 0.2089,
+      "step": 1425
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5060101747512817,
+      "learning_rate": 2.513309416885865e-05,
+      "loss": 0.1674,
+      "step": 1426
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6767317652702332,
+      "learning_rate": 2.5066517691603324e-05,
+      "loss": 0.2047,
+      "step": 1427
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.7371398210525513,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 0.2757,
+      "step": 1428
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5077962279319763,
+      "learning_rate": 2.493354125087738e-05,
+      "loss": 0.1668,
+      "step": 1429
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5125499963760376,
+      "learning_rate": 2.4867141600925214e-05,
+      "loss": 0.1937,
+      "step": 1430
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.41824251413345337,
+      "learning_rate": 2.4800801206693875e-05,
+      "loss": 0.1294,
+      "step": 1431
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.3940209448337555,
+      "learning_rate": 2.4734520224594093e-05,
+      "loss": 0.1326,
+      "step": 1432
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5067299604415894,
+      "learning_rate": 2.4668298810896462e-05,
+      "loss": 0.1942,
+      "step": 1433
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5119947195053101,
+      "learning_rate": 2.4602137121731195e-05,
+      "loss": 0.1361,
+      "step": 1434
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.4600105285644531,
+      "learning_rate": 2.4536035313087602e-05,
+      "loss": 0.1375,
+      "step": 1435
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5646739602088928,
+      "learning_rate": 2.44699935408139e-05,
+      "loss": 0.192,
+      "step": 1436
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.49743449687957764,
+      "learning_rate": 2.4404011960616747e-05,
+      "loss": 0.1623,
+      "step": 1437
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5839107036590576,
+      "learning_rate": 2.433809072806081e-05,
+      "loss": 0.2112,
+      "step": 1438
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5835247039794922,
+      "learning_rate": 2.4272229998568575e-05,
+      "loss": 0.2285,
+      "step": 1439
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5345538258552551,
+      "learning_rate": 2.4206429927419794e-05,
+      "loss": 0.1824,
+      "step": 1440
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6125417947769165,
+      "learning_rate": 2.4140690669751283e-05,
+      "loss": 0.2012,
+      "step": 1441
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6538706421852112,
+      "learning_rate": 2.40750123805564e-05,
+      "loss": 0.2031,
+      "step": 1442
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.4648752808570862,
+      "learning_rate": 2.4009395214684838e-05,
+      "loss": 0.1628,
+      "step": 1443
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6477990746498108,
+      "learning_rate": 2.3943839326842092e-05,
+      "loss": 0.1977,
+      "step": 1444
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6007817983627319,
+      "learning_rate": 2.3878344871589257e-05,
+      "loss": 0.2008,
+      "step": 1445
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5213921666145325,
+      "learning_rate": 2.3812912003342568e-05,
+      "loss": 0.1562,
+      "step": 1446
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6278839707374573,
+      "learning_rate": 2.3747540876373025e-05,
+      "loss": 0.2188,
+      "step": 1447
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5474607348442078,
+      "learning_rate": 2.3682231644806108e-05,
+      "loss": 0.178,
+      "step": 1448
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6178276538848877,
+      "learning_rate": 2.3616984462621305e-05,
+      "loss": 0.1922,
+      "step": 1449
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.580036997795105,
+      "learning_rate": 2.3551799483651894e-05,
+      "loss": 0.1646,
+      "step": 1450
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5910034775733948,
+      "learning_rate": 2.3486676861584405e-05,
+      "loss": 0.2174,
+      "step": 1451
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5000534653663635,
+      "learning_rate": 2.342161674995843e-05,
+      "loss": 0.1715,
+      "step": 1452
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.49306708574295044,
+      "learning_rate": 2.335661930216611e-05,
+      "loss": 0.1649,
+      "step": 1453
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5357517004013062,
+      "learning_rate": 2.3291684671451903e-05,
+      "loss": 0.2108,
+      "step": 1454
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5509576201438904,
+      "learning_rate": 2.322681301091214e-05,
+      "loss": 0.2064,
+      "step": 1455
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5236726403236389,
+      "learning_rate": 2.3162004473494657e-05,
+      "loss": 0.1353,
+      "step": 1456
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.49893948435783386,
+      "learning_rate": 2.3097259211998534e-05,
+      "loss": 0.1445,
+      "step": 1457
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5097013711929321,
+      "learning_rate": 2.3032577379073577e-05,
+      "loss": 0.1465,
+      "step": 1458
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.435596764087677,
+      "learning_rate": 2.296795912722014e-05,
+      "loss": 0.1291,
+      "step": 1459
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5503084659576416,
+      "learning_rate": 2.2903404608788582e-05,
+      "loss": 0.1209,
+      "step": 1460
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.6414248943328857,
+      "learning_rate": 2.283891397597908e-05,
+      "loss": 0.2211,
+      "step": 1461
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5442419052124023,
+      "learning_rate": 2.2774487380841115e-05,
+      "loss": 0.2035,
+      "step": 1462
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.609218955039978,
+      "learning_rate": 2.2710124975273234e-05,
+      "loss": 0.2328,
+      "step": 1463
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5088512897491455,
+      "learning_rate": 2.2645826911022655e-05,
+      "loss": 0.1633,
+      "step": 1464
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.6089456677436829,
+      "learning_rate": 2.2581593339684835e-05,
+      "loss": 0.1424,
+      "step": 1465
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.6338807344436646,
+      "learning_rate": 2.2517424412703252e-05,
+      "loss": 0.2143,
+      "step": 1466
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.6188156604766846,
+      "learning_rate": 2.24533202813689e-05,
+      "loss": 0.189,
+      "step": 1467
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.630052387714386,
+      "learning_rate": 2.2389281096820075e-05,
+      "loss": 0.1874,
+      "step": 1468
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5420491099357605,
+      "learning_rate": 2.2325307010041874e-05,
+      "loss": 0.1562,
+      "step": 1469
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.4673231840133667,
+      "learning_rate": 2.2261398171865978e-05,
+      "loss": 0.1373,
+      "step": 1470
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.4499804377555847,
+      "learning_rate": 2.21975547329702e-05,
+      "loss": 0.1467,
+      "step": 1471
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.6852113604545593,
+      "learning_rate": 2.2133776843878186e-05,
+      "loss": 0.2925,
+      "step": 1472
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.7545210719108582,
+      "learning_rate": 2.2070064654958976e-05,
+      "loss": 0.2723,
+      "step": 1473
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5475196838378906,
+      "learning_rate": 2.2006418316426775e-05,
+      "loss": 0.1859,
+      "step": 1474
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5056906938552856,
+      "learning_rate": 2.1942837978340514e-05,
+      "loss": 0.1604,
+      "step": 1475
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.4816121757030487,
+      "learning_rate": 2.187932379060348e-05,
+      "loss": 0.1619,
+      "step": 1476
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5058569312095642,
+      "learning_rate": 2.1815875902963055e-05,
+      "loss": 0.1619,
+      "step": 1477
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5065352916717529,
+      "learning_rate": 2.1752494465010242e-05,
+      "loss": 0.1829,
+      "step": 1478
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5500874519348145,
+      "learning_rate": 2.168917962617944e-05,
+      "loss": 0.1587,
+      "step": 1479
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6841371655464172,
+      "learning_rate": 2.1625931535747964e-05,
+      "loss": 0.2459,
+      "step": 1480
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5658700466156006,
+      "learning_rate": 2.1562750342835826e-05,
+      "loss": 0.1664,
+      "step": 1481
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.7092359662055969,
+      "learning_rate": 2.1499636196405228e-05,
+      "loss": 0.1979,
+      "step": 1482
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.547674298286438,
+      "learning_rate": 2.1436589245260376e-05,
+      "loss": 0.1358,
+      "step": 1483
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.7069271802902222,
+      "learning_rate": 2.1373609638047032e-05,
+      "loss": 0.2189,
+      "step": 1484
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.7143983244895935,
+      "learning_rate": 2.1310697523252127e-05,
+      "loss": 0.2739,
+      "step": 1485
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6037214398384094,
+      "learning_rate": 2.1247853049203543e-05,
+      "loss": 0.2093,
+      "step": 1486
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5048996806144714,
+      "learning_rate": 2.1185076364069617e-05,
+      "loss": 0.1867,
+      "step": 1487
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6051159501075745,
+      "learning_rate": 2.112236761585892e-05,
+      "loss": 0.2115,
+      "step": 1488
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.711506724357605,
+      "learning_rate": 2.1059726952419782e-05,
+      "loss": 0.2853,
+      "step": 1489
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6495888233184814,
+      "learning_rate": 2.09971545214401e-05,
+      "loss": 0.2706,
+      "step": 1490
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.4984496235847473,
+      "learning_rate": 2.0934650470446786e-05,
+      "loss": 0.1725,
+      "step": 1491
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5288287997245789,
+      "learning_rate": 2.087221494680563e-05,
+      "loss": 0.155,
+      "step": 1492
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6655897498130798,
+      "learning_rate": 2.0809848097720826e-05,
+      "loss": 0.261,
+      "step": 1493
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.4433102309703827,
+      "learning_rate": 2.074755007023461e-05,
+      "loss": 0.1293,
+      "step": 1494
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5946065783500671,
+      "learning_rate": 2.0685321011227036e-05,
+      "loss": 0.1926,
+      "step": 1495
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5296228528022766,
+      "learning_rate": 2.0623161067415463e-05,
+      "loss": 0.2027,
+      "step": 1496
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5599090456962585,
+      "learning_rate": 2.0561070385354387e-05,
+      "loss": 0.1491,
+      "step": 1497
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.6422383189201355,
+      "learning_rate": 2.049904911143492e-05,
+      "loss": 0.2079,
+      "step": 1498
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.4741975665092468,
+      "learning_rate": 2.0437097391884614e-05,
+      "loss": 0.1337,
+      "step": 1499
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.6488420367240906,
+      "learning_rate": 2.0375215372766944e-05,
+      "loss": 0.2497,
+      "step": 1500
+    },
+    {
+      "epoch": 1.42,
+      "eval_loss": 0.29373255372047424,
+      "eval_runtime": 365.1223,
+      "eval_samples_per_second": 0.934,
+      "eval_steps_per_second": 0.236,
+      "step": 1500
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7022296786308289,
+      "learning_rate": 2.0313403199981124e-05,
+      "loss": 0.2604,
+      "step": 1501
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5766379237174988,
+      "learning_rate": 2.025166101926168e-05,
+      "loss": 0.1703,
+      "step": 1502
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7454899549484253,
+      "learning_rate": 2.0189988976178077e-05,
+      "loss": 0.2695,
+      "step": 1503
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5942023992538452,
+      "learning_rate": 2.012838721613447e-05,
+      "loss": 0.1977,
+      "step": 1504
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.6005353927612305,
+      "learning_rate": 2.0066855884369244e-05,
+      "loss": 0.2011,
+      "step": 1505
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.6216517090797424,
+      "learning_rate": 2.0005395125954812e-05,
+      "loss": 0.1802,
+      "step": 1506
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5090690851211548,
+      "learning_rate": 1.9944005085797124e-05,
+      "loss": 0.1469,
+      "step": 1507
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.713166356086731,
+      "learning_rate": 1.9882685908635462e-05,
+      "loss": 0.2004,
+      "step": 1508
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.6232782006263733,
+      "learning_rate": 1.982143773904197e-05,
+      "loss": 0.1953,
+      "step": 1509
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.548896312713623,
+      "learning_rate": 1.9760260721421426e-05,
+      "loss": 0.1882,
+      "step": 1510
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.41742146015167236,
+      "learning_rate": 1.9699155000010855e-05,
+      "loss": 0.1334,
+      "step": 1511
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5797178149223328,
+      "learning_rate": 1.963812071887913e-05,
+      "loss": 0.1769,
+      "step": 1512
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5724051594734192,
+      "learning_rate": 1.9577158021926774e-05,
+      "loss": 0.1867,
+      "step": 1513
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5145040154457092,
+      "learning_rate": 1.951626705288544e-05,
+      "loss": 0.1728,
+      "step": 1514
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6977815628051758,
+      "learning_rate": 1.9455447955317767e-05,
+      "loss": 0.2346,
+      "step": 1515
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.46963533759117126,
+      "learning_rate": 1.9394700872616855e-05,
+      "loss": 0.1214,
+      "step": 1516
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6454631686210632,
+      "learning_rate": 1.9334025948006074e-05,
+      "loss": 0.2133,
+      "step": 1517
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.7710499167442322,
+      "learning_rate": 1.9273423324538658e-05,
+      "loss": 0.3188,
+      "step": 1518
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.5848934054374695,
+      "learning_rate": 1.9212893145097338e-05,
+      "loss": 0.237,
+      "step": 1519
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.48352962732315063,
+      "learning_rate": 1.9152435552394104e-05,
+      "loss": 0.128,
+      "step": 1520
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.540931224822998,
+      "learning_rate": 1.9092050688969738e-05,
+      "loss": 0.1983,
+      "step": 1521
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.47986218333244324,
+      "learning_rate": 1.903173869719362e-05,
+      "loss": 0.1357,
+      "step": 1522
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.5026472806930542,
+      "learning_rate": 1.8971499719263252e-05,
+      "loss": 0.167,
+      "step": 1523
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.462436318397522,
+      "learning_rate": 1.891133389720407e-05,
+      "loss": 0.1465,
+      "step": 1524
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.5045827627182007,
+      "learning_rate": 1.8851241372868937e-05,
+      "loss": 0.1657,
+      "step": 1525
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.4786836504936218,
+      "learning_rate": 1.8791222287937982e-05,
+      "loss": 0.1816,
+      "step": 1526
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.7790905833244324,
+      "learning_rate": 1.873127678391816e-05,
+      "loss": 0.3094,
+      "step": 1527
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5625799298286438,
+      "learning_rate": 1.8671405002142918e-05,
+      "loss": 0.2455,
+      "step": 1528
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.6317288279533386,
+      "learning_rate": 1.861160708377193e-05,
+      "loss": 0.2116,
+      "step": 1529
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5869625210762024,
+      "learning_rate": 1.855188316979068e-05,
+      "loss": 0.2176,
+      "step": 1530
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5973176956176758,
+      "learning_rate": 1.8492233401010218e-05,
+      "loss": 0.1976,
+      "step": 1531
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.500684916973114,
+      "learning_rate": 1.8432657918066732e-05,
+      "loss": 0.1405,
+      "step": 1532
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.4859834909439087,
+      "learning_rate": 1.8373156861421325e-05,
+      "loss": 0.1591,
+      "step": 1533
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.48588091135025024,
+      "learning_rate": 1.831373037135955e-05,
+      "loss": 0.2044,
+      "step": 1534
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.6047201752662659,
+      "learning_rate": 1.8254378587991226e-05,
+      "loss": 0.2231,
+      "step": 1535
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5573422312736511,
+      "learning_rate": 1.8195101651250017e-05,
+      "loss": 0.1597,
+      "step": 1536
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.6479578018188477,
+      "learning_rate": 1.8135899700893076e-05,
+      "loss": 0.2469,
+      "step": 1537
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5928018689155579,
+      "learning_rate": 1.807677287650083e-05,
+      "loss": 0.2051,
+      "step": 1538
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5605274438858032,
+      "learning_rate": 1.8017721317476517e-05,
+      "loss": 0.1925,
+      "step": 1539
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5107634663581848,
+      "learning_rate": 1.7958745163045986e-05,
+      "loss": 0.1832,
+      "step": 1540
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.42251497507095337,
+      "learning_rate": 1.7899844552257233e-05,
+      "loss": 0.1265,
+      "step": 1541
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5432945489883423,
+      "learning_rate": 1.7841019623980217e-05,
+      "loss": 0.1841,
+      "step": 1542
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.6625102758407593,
+      "learning_rate": 1.778227051690639e-05,
+      "loss": 0.172,
+      "step": 1543
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.6276701092720032,
+      "learning_rate": 1.77235973695485e-05,
+      "loss": 0.1511,
+      "step": 1544
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.5530522465705872,
+      "learning_rate": 1.76650003202402e-05,
+      "loss": 0.2513,
+      "step": 1545
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.48519518971443176,
+      "learning_rate": 1.760647950713566e-05,
+      "loss": 0.1616,
+      "step": 1546
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.6004926562309265,
+      "learning_rate": 1.75480350682094e-05,
+      "loss": 0.2303,
+      "step": 1547
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5747936367988586,
+      "learning_rate": 1.74896671412558e-05,
+      "loss": 0.2214,
+      "step": 1548
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5651929378509521,
+      "learning_rate": 1.7431375863888898e-05,
+      "loss": 0.2182,
+      "step": 1549
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.6708860397338867,
+      "learning_rate": 1.737316137354197e-05,
+      "loss": 0.2916,
+      "step": 1550
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.439681738615036,
+      "learning_rate": 1.7315023807467295e-05,
+      "loss": 0.1117,
+      "step": 1551
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5426954030990601,
+      "learning_rate": 1.725696330273575e-05,
+      "loss": 0.2089,
+      "step": 1552
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.44387537240982056,
+      "learning_rate": 1.7198979996236546e-05,
+      "loss": 0.1546,
+      "step": 1553
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5127511620521545,
+      "learning_rate": 1.7141074024676912e-05,
+      "loss": 0.1656,
+      "step": 1554
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5790517330169678,
+      "learning_rate": 1.7083245524581666e-05,
+      "loss": 0.2073,
+      "step": 1555
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5386837124824524,
+      "learning_rate": 1.7025494632293048e-05,
+      "loss": 0.1714,
+      "step": 1556
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.382916659116745,
+      "learning_rate": 1.6967821483970277e-05,
+      "loss": 0.079,
+      "step": 1557
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.48617151379585266,
+      "learning_rate": 1.6910226215589303e-05,
+      "loss": 0.1835,
+      "step": 1558
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.42700323462486267,
+      "learning_rate": 1.6852708962942427e-05,
+      "loss": 0.14,
+      "step": 1559
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.63298100233078,
+      "learning_rate": 1.6795269861638042e-05,
+      "loss": 0.2174,
+      "step": 1560
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6499307751655579,
+      "learning_rate": 1.6737909047100293e-05,
+      "loss": 0.246,
+      "step": 1561
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.5038250684738159,
+      "learning_rate": 1.6680626654568686e-05,
+      "loss": 0.1535,
+      "step": 1562
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.4813094735145569,
+      "learning_rate": 1.6623422819097917e-05,
+      "loss": 0.1528,
+      "step": 1563
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.677268385887146,
+      "learning_rate": 1.656629767555739e-05,
+      "loss": 0.1932,
+      "step": 1564
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6332627534866333,
+      "learning_rate": 1.650925135863104e-05,
+      "loss": 0.2713,
+      "step": 1565
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.5338670611381531,
+      "learning_rate": 1.645228400281689e-05,
+      "loss": 0.1559,
+      "step": 1566
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6017856597900391,
+      "learning_rate": 1.6395395742426874e-05,
+      "loss": 0.2186,
+      "step": 1567
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.5935778617858887,
+      "learning_rate": 1.6338586711586356e-05,
+      "loss": 0.221,
+      "step": 1568
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.4849663972854614,
+      "learning_rate": 1.6281857044233968e-05,
+      "loss": 0.1664,
+      "step": 1569
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.6336452960968018,
+      "learning_rate": 1.6225206874121218e-05,
+      "loss": 0.1974,
+      "step": 1570
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.4798564016819,
+      "learning_rate": 1.6168636334812125e-05,
+      "loss": 0.1308,
+      "step": 1571
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.5006871819496155,
+      "learning_rate": 1.6112145559683055e-05,
+      "loss": 0.1972,
+      "step": 1572
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.30803796648979187,
+      "learning_rate": 1.6055734681922223e-05,
+      "loss": 0.0783,
+      "step": 1573
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.647851824760437,
+      "learning_rate": 1.5999403834529548e-05,
+      "loss": 0.2216,
+      "step": 1574
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.597584068775177,
+      "learning_rate": 1.5943153150316193e-05,
+      "loss": 0.2061,
+      "step": 1575
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.5646488666534424,
+      "learning_rate": 1.5886982761904377e-05,
+      "loss": 0.1601,
+      "step": 1576
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.5062416791915894,
+      "learning_rate": 1.583089280172696e-05,
+      "loss": 0.1572,
+      "step": 1577
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5994753837585449,
+      "learning_rate": 1.5774883402027208e-05,
+      "loss": 0.2155,
+      "step": 1578
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.48443084955215454,
+      "learning_rate": 1.5718954694858458e-05,
+      "loss": 0.1537,
+      "step": 1579
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6129524111747742,
+      "learning_rate": 1.5663106812083743e-05,
+      "loss": 0.2066,
+      "step": 1580
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5396411418914795,
+      "learning_rate": 1.5607339885375617e-05,
+      "loss": 0.154,
+      "step": 1581
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6866434812545776,
+      "learning_rate": 1.555165404621567e-05,
+      "loss": 0.2622,
+      "step": 1582
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5122365355491638,
+      "learning_rate": 1.549604942589441e-05,
+      "loss": 0.1686,
+      "step": 1583
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5591203570365906,
+      "learning_rate": 1.5440526155510763e-05,
+      "loss": 0.2545,
+      "step": 1584
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5799269676208496,
+      "learning_rate": 1.5385084365971948e-05,
+      "loss": 0.224,
+      "step": 1585
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.4950427711009979,
+      "learning_rate": 1.532972418799298e-05,
+      "loss": 0.1892,
+      "step": 1586
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.4790683090686798,
+      "learning_rate": 1.527444575209654e-05,
+      "loss": 0.1707,
+      "step": 1587
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.49954721331596375,
+      "learning_rate": 1.5219249188612556e-05,
+      "loss": 0.2106,
+      "step": 1588
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5916127562522888,
+      "learning_rate": 1.5164134627677895e-05,
+      "loss": 0.2768,
+      "step": 1589
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5299326777458191,
+      "learning_rate": 1.5109102199236152e-05,
+      "loss": 0.221,
+      "step": 1590
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5694025158882141,
+      "learning_rate": 1.5054152033037205e-05,
+      "loss": 0.2141,
+      "step": 1591
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5819464325904846,
+      "learning_rate": 1.4999284258637053e-05,
+      "loss": 0.1539,
+      "step": 1592
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5930538177490234,
+      "learning_rate": 1.4944499005397371e-05,
+      "loss": 0.2458,
+      "step": 1593
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.5449967384338379,
+      "learning_rate": 1.488979640248534e-05,
+      "loss": 0.2011,
+      "step": 1594
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.47328758239746094,
+      "learning_rate": 1.4835176578873212e-05,
+      "loss": 0.1452,
+      "step": 1595
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.48782655596733093,
+      "learning_rate": 1.4780639663338125e-05,
+      "loss": 0.1474,
+      "step": 1596
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.45278820395469666,
+      "learning_rate": 1.4726185784461728e-05,
+      "loss": 0.1203,
+      "step": 1597
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.553763747215271,
+      "learning_rate": 1.4671815070629868e-05,
+      "loss": 0.2043,
+      "step": 1598
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5404416918754578,
+      "learning_rate": 1.461752765003236e-05,
+      "loss": 0.145,
+      "step": 1599
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5086386799812317,
+      "learning_rate": 1.4563323650662586e-05,
+      "loss": 0.1789,
+      "step": 1600
+    },
+    {
+      "epoch": 1.52,
+      "eval_loss": 0.2908007502555847,
+      "eval_runtime": 374.4997,
+      "eval_samples_per_second": 0.911,
+      "eval_steps_per_second": 0.23,
+      "step": 1600
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.43449389934539795,
+      "learning_rate": 1.45092032003173e-05,
+      "loss": 0.1151,
+      "step": 1601
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5409952998161316,
+      "learning_rate": 1.4455166426596223e-05,
+      "loss": 0.1845,
+      "step": 1602
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.4433218538761139,
+      "learning_rate": 1.4401213456901819e-05,
+      "loss": 0.1506,
+      "step": 1603
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.6740498542785645,
+      "learning_rate": 1.434734441843899e-05,
+      "loss": 0.164,
+      "step": 1604
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.4789322018623352,
+      "learning_rate": 1.4293559438214688e-05,
+      "loss": 0.1622,
+      "step": 1605
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5942570567131042,
+      "learning_rate": 1.4239858643037751e-05,
+      "loss": 0.1602,
+      "step": 1606
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.6144424080848694,
+      "learning_rate": 1.4186242159518476e-05,
+      "loss": 0.1696,
+      "step": 1607
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5805176496505737,
+      "learning_rate": 1.4132710114068426e-05,
+      "loss": 0.1553,
+      "step": 1608
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.38980087637901306,
+      "learning_rate": 1.4079262632900047e-05,
+      "loss": 0.0943,
+      "step": 1609
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.564573347568512,
+      "learning_rate": 1.4025899842026442e-05,
+      "loss": 0.1808,
+      "step": 1610
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6322575807571411,
+      "learning_rate": 1.3972621867260998e-05,
+      "loss": 0.2339,
+      "step": 1611
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5197539925575256,
+      "learning_rate": 1.3919428834217163e-05,
+      "loss": 0.1832,
+      "step": 1612
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.48535871505737305,
+      "learning_rate": 1.3866320868308136e-05,
+      "loss": 0.1158,
+      "step": 1613
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5294753313064575,
+      "learning_rate": 1.3813298094746491e-05,
+      "loss": 0.1435,
+      "step": 1614
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6298293471336365,
+      "learning_rate": 1.3760360638544012e-05,
+      "loss": 0.1968,
+      "step": 1615
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5538610816001892,
+      "learning_rate": 1.3707508624511262e-05,
+      "loss": 0.1723,
+      "step": 1616
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6312843561172485,
+      "learning_rate": 1.3654742177257434e-05,
+      "loss": 0.1702,
+      "step": 1617
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6342264413833618,
+      "learning_rate": 1.36020614211899e-05,
+      "loss": 0.2414,
+      "step": 1618
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6629549264907837,
+      "learning_rate": 1.3549466480514078e-05,
+      "loss": 0.2058,
+      "step": 1619
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5926294326782227,
+      "learning_rate": 1.349695747923298e-05,
+      "loss": 0.1811,
+      "step": 1620
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5224953889846802,
+      "learning_rate": 1.3444534541147057e-05,
+      "loss": 0.1535,
+      "step": 1621
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.44833263754844666,
+      "learning_rate": 1.339219778985385e-05,
+      "loss": 0.1365,
+      "step": 1622
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5778658390045166,
+      "learning_rate": 1.3339947348747633e-05,
+      "loss": 0.1797,
+      "step": 1623
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5402314066886902,
+      "learning_rate": 1.3287783341019278e-05,
+      "loss": 0.1646,
+      "step": 1624
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5998130440711975,
+      "learning_rate": 1.323570588965578e-05,
+      "loss": 0.199,
+      "step": 1625
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5934169888496399,
+      "learning_rate": 1.3183715117440142e-05,
+      "loss": 0.1922,
+      "step": 1626
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.7084951400756836,
+      "learning_rate": 1.3131811146950945e-05,
+      "loss": 0.2118,
+      "step": 1627
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5603919625282288,
+      "learning_rate": 1.3079994100562159e-05,
+      "loss": 0.1882,
+      "step": 1628
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.3938945233821869,
+      "learning_rate": 1.3028264100442771e-05,
+      "loss": 0.1126,
+      "step": 1629
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.550576388835907,
+      "learning_rate": 1.2976621268556571e-05,
+      "loss": 0.1947,
+      "step": 1630
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5770824551582336,
+      "learning_rate": 1.2925065726661844e-05,
+      "loss": 0.1521,
+      "step": 1631
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6677595973014832,
+      "learning_rate": 1.2873597596311027e-05,
+      "loss": 0.2037,
+      "step": 1632
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.48562943935394287,
+      "learning_rate": 1.2822216998850506e-05,
+      "loss": 0.1691,
+      "step": 1633
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5756161212921143,
+      "learning_rate": 1.2770924055420258e-05,
+      "loss": 0.1754,
+      "step": 1634
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.4862710237503052,
+      "learning_rate": 1.2719718886953646e-05,
+      "loss": 0.1865,
+      "step": 1635
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.693092405796051,
+      "learning_rate": 1.2668601614177017e-05,
+      "loss": 0.2322,
+      "step": 1636
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5858150720596313,
+      "learning_rate": 1.2617572357609564e-05,
+      "loss": 0.2106,
+      "step": 1637
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6116892695426941,
+      "learning_rate": 1.2566631237562893e-05,
+      "loss": 0.1982,
+      "step": 1638
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.505799412727356,
+      "learning_rate": 1.2515778374140858e-05,
+      "loss": 0.187,
+      "step": 1639
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5432391166687012,
+      "learning_rate": 1.246501388723923e-05,
+      "loss": 0.197,
+      "step": 1640
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.45380833745002747,
+      "learning_rate": 1.2414337896545374e-05,
+      "loss": 0.1198,
+      "step": 1641
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.4624657928943634,
+      "learning_rate": 1.2363750521538064e-05,
+      "loss": 0.1274,
+      "step": 1642
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5638636350631714,
+      "learning_rate": 1.2313251881487082e-05,
+      "loss": 0.2107,
+      "step": 1643
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5158945918083191,
+      "learning_rate": 1.2262842095453065e-05,
+      "loss": 0.1422,
+      "step": 1644
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.6319577097892761,
+      "learning_rate": 1.2212521282287092e-05,
+      "loss": 0.1966,
+      "step": 1645
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5805641412734985,
+      "learning_rate": 1.2162289560630524e-05,
+      "loss": 0.1704,
+      "step": 1646
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5535603165626526,
+      "learning_rate": 1.2112147048914668e-05,
+      "loss": 0.1653,
+      "step": 1647
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5752933621406555,
+      "learning_rate": 1.2062093865360458e-05,
+      "loss": 0.2224,
+      "step": 1648
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.49497225880622864,
+      "learning_rate": 1.2012130127978266e-05,
+      "loss": 0.1797,
+      "step": 1649
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5936898589134216,
+      "learning_rate": 1.1962255954567537e-05,
+      "loss": 0.2252,
+      "step": 1650
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5926779508590698,
+      "learning_rate": 1.1912471462716595e-05,
+      "loss": 0.2359,
+      "step": 1651
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.6608781218528748,
+      "learning_rate": 1.1862776769802276e-05,
+      "loss": 0.2373,
+      "step": 1652
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.4078502357006073,
+      "learning_rate": 1.1813171992989741e-05,
+      "loss": 0.1107,
+      "step": 1653
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.45365238189697266,
+      "learning_rate": 1.1763657249232107e-05,
+      "loss": 0.16,
+      "step": 1654
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.6005438566207886,
+      "learning_rate": 1.1714232655270263e-05,
+      "loss": 0.1856,
+      "step": 1655
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5458640456199646,
+      "learning_rate": 1.1664898327632551e-05,
+      "loss": 0.208,
+      "step": 1656
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.46168598532676697,
+      "learning_rate": 1.1615654382634444e-05,
+      "loss": 0.1601,
+      "step": 1657
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.466379851102829,
+      "learning_rate": 1.1566500936378388e-05,
+      "loss": 0.1314,
+      "step": 1658
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5661909580230713,
+      "learning_rate": 1.1517438104753386e-05,
+      "loss": 0.2149,
+      "step": 1659
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.49127811193466187,
+      "learning_rate": 1.146846600343488e-05,
+      "loss": 0.1475,
+      "step": 1660
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5873091220855713,
+      "learning_rate": 1.1419584747884322e-05,
+      "loss": 0.2238,
+      "step": 1661
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.4300805330276489,
+      "learning_rate": 1.1370794453349037e-05,
+      "loss": 0.1141,
+      "step": 1662
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.5848353505134583,
+      "learning_rate": 1.132209523486184e-05,
+      "loss": 0.1845,
+      "step": 1663
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.5535120368003845,
+      "learning_rate": 1.1273487207240847e-05,
+      "loss": 0.1887,
+      "step": 1664
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4587995707988739,
+      "learning_rate": 1.1224970485089192e-05,
+      "loss": 0.127,
+      "step": 1665
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4950588047504425,
+      "learning_rate": 1.1176545182794674e-05,
+      "loss": 0.1458,
+      "step": 1666
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.5891401171684265,
+      "learning_rate": 1.1128211414529627e-05,
+      "loss": 0.2055,
+      "step": 1667
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.5351936221122742,
+      "learning_rate": 1.1079969294250514e-05,
+      "loss": 0.1988,
+      "step": 1668
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.38483020663261414,
+      "learning_rate": 1.1031818935697763e-05,
+      "loss": 0.0834,
+      "step": 1669
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4198475778102875,
+      "learning_rate": 1.0983760452395413e-05,
+      "loss": 0.1078,
+      "step": 1670
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.5883978009223938,
+      "learning_rate": 1.0935793957650948e-05,
+      "loss": 0.1187,
+      "step": 1671
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4993918836116791,
+      "learning_rate": 1.0887919564554894e-05,
+      "loss": 0.1416,
+      "step": 1672
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.570749819278717,
+      "learning_rate": 1.0840137385980697e-05,
+      "loss": 0.147,
+      "step": 1673
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.5229507088661194,
+      "learning_rate": 1.079244753458437e-05,
+      "loss": 0.1782,
+      "step": 1674
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6603185534477234,
+      "learning_rate": 1.0744850122804218e-05,
+      "loss": 0.2184,
+      "step": 1675
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.594048261642456,
+      "learning_rate": 1.0697345262860636e-05,
+      "loss": 0.2123,
+      "step": 1676
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6253306269645691,
+      "learning_rate": 1.064993306675578e-05,
+      "loss": 0.2096,
+      "step": 1677
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.5980146527290344,
+      "learning_rate": 1.0602613646273374e-05,
+      "loss": 0.1241,
+      "step": 1678
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.49638286232948303,
+      "learning_rate": 1.0555387112978349e-05,
+      "loss": 0.1414,
+      "step": 1679
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.468610554933548,
+      "learning_rate": 1.0508253578216693e-05,
+      "loss": 0.1364,
+      "step": 1680
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6512147188186646,
+      "learning_rate": 1.046121315311508e-05,
+      "loss": 0.2184,
+      "step": 1681
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.44908082485198975,
+      "learning_rate": 1.0414265948580692e-05,
+      "loss": 0.1415,
+      "step": 1682
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.5168013572692871,
+      "learning_rate": 1.0367412075300942e-05,
+      "loss": 0.1355,
+      "step": 1683
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.4128986597061157,
+      "learning_rate": 1.032065164374313e-05,
+      "loss": 0.1204,
+      "step": 1684
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.4346331059932709,
+      "learning_rate": 1.0273984764154326e-05,
+      "loss": 0.1219,
+      "step": 1685
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.48234665393829346,
+      "learning_rate": 1.0227411546560962e-05,
+      "loss": 0.1525,
+      "step": 1686
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.672383189201355,
+      "learning_rate": 1.0180932100768715e-05,
+      "loss": 0.1578,
+      "step": 1687
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.48131534457206726,
+      "learning_rate": 1.01345465363621e-05,
+      "loss": 0.1525,
+      "step": 1688
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.4903189539909363,
+      "learning_rate": 1.0088254962704341e-05,
+      "loss": 0.1351,
+      "step": 1689
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5856617093086243,
+      "learning_rate": 1.0042057488937067e-05,
+      "loss": 0.1786,
+      "step": 1690
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.6022653579711914,
+      "learning_rate": 9.995954223979992e-06,
+      "loss": 0.2025,
+      "step": 1691
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5132540464401245,
+      "learning_rate": 9.949945276530781e-06,
+      "loss": 0.1581,
+      "step": 1692
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.6871017813682556,
+      "learning_rate": 9.904030755064659e-06,
+      "loss": 0.2072,
+      "step": 1693
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.47669410705566406,
+      "learning_rate": 9.858210767834292e-06,
+      "loss": 0.1535,
+      "step": 1694
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5964915156364441,
+      "learning_rate": 9.8124854228694e-06,
+      "loss": 0.2018,
+      "step": 1695
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.4954380393028259,
+      "learning_rate": 9.766854827976617e-06,
+      "loss": 0.1559,
+      "step": 1696
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.524189293384552,
+      "learning_rate": 9.72131909073914e-06,
+      "loss": 0.1634,
+      "step": 1697
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.6875957250595093,
+      "learning_rate": 9.675878318516546e-06,
+      "loss": 0.2073,
+      "step": 1698
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5479033589363098,
+      "learning_rate": 9.630532618444533e-06,
+      "loss": 0.1216,
+      "step": 1699
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.6077048778533936,
+      "learning_rate": 9.58528209743459e-06,
+      "loss": 0.1853,
+      "step": 1700
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 0.2951177656650543,
+      "eval_runtime": 379.6741,
+      "eval_samples_per_second": 0.898,
+      "eval_steps_per_second": 0.227,
+      "step": 1700
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5721592903137207,
+      "learning_rate": 9.540126862173864e-06,
+      "loss": 0.2119,
+      "step": 1701
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.6499807834625244,
+      "learning_rate": 9.495067019124792e-06,
+      "loss": 0.215,
+      "step": 1702
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5961745381355286,
+      "learning_rate": 9.450102674524952e-06,
+      "loss": 0.1981,
+      "step": 1703
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5602648854255676,
+      "learning_rate": 9.405233934386726e-06,
+      "loss": 0.167,
+      "step": 1704
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5367100834846497,
+      "learning_rate": 9.360460904497131e-06,
+      "loss": 0.1478,
+      "step": 1705
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5618895888328552,
+      "learning_rate": 9.315783690417479e-06,
+      "loss": 0.1871,
+      "step": 1706
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.6565399169921875,
+      "learning_rate": 9.271202397483215e-06,
+      "loss": 0.2341,
+      "step": 1707
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.43295586109161377,
+      "learning_rate": 9.226717130803636e-06,
+      "loss": 0.1394,
+      "step": 1708
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.7067288756370544,
+      "learning_rate": 9.182327995261591e-06,
+      "loss": 0.237,
+      "step": 1709
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5122333765029907,
+      "learning_rate": 9.138035095513336e-06,
+      "loss": 0.1395,
+      "step": 1710
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5410517454147339,
+      "learning_rate": 9.09383853598818e-06,
+      "loss": 0.1281,
+      "step": 1711
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.6178551912307739,
+      "learning_rate": 9.049738420888349e-06,
+      "loss": 0.1849,
+      "step": 1712
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5236595869064331,
+      "learning_rate": 9.005734854188625e-06,
+      "loss": 0.1553,
+      "step": 1713
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5782197713851929,
+      "learning_rate": 8.961827939636196e-06,
+      "loss": 0.1912,
+      "step": 1714
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.5927847623825073,
+      "learning_rate": 8.918017780750349e-06,
+      "loss": 0.1733,
+      "step": 1715
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.6102494597434998,
+      "learning_rate": 8.87430448082227e-06,
+      "loss": 0.1757,
+      "step": 1716
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.628868579864502,
+      "learning_rate": 8.830688142914784e-06,
+      "loss": 0.2016,
+      "step": 1717
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.45962807536125183,
+      "learning_rate": 8.787168869862066e-06,
+      "loss": 0.1451,
+      "step": 1718
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.6048905253410339,
+      "learning_rate": 8.743746764269512e-06,
+      "loss": 0.2302,
+      "step": 1719
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.6292806267738342,
+      "learning_rate": 8.700421928513352e-06,
+      "loss": 0.1546,
+      "step": 1720
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.49050647020339966,
+      "learning_rate": 8.657194464740542e-06,
+      "loss": 0.1713,
+      "step": 1721
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.6007164120674133,
+      "learning_rate": 8.614064474868421e-06,
+      "loss": 0.1799,
+      "step": 1722
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.49299028515815735,
+      "learning_rate": 8.571032060584556e-06,
+      "loss": 0.1665,
+      "step": 1723
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.4100872874259949,
+      "learning_rate": 8.528097323346407e-06,
+      "loss": 0.1162,
+      "step": 1724
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.6367151141166687,
+      "learning_rate": 8.485260364381187e-06,
+      "loss": 0.2045,
+      "step": 1725
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5315971970558167,
+      "learning_rate": 8.442521284685573e-06,
+      "loss": 0.1472,
+      "step": 1726
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5666583776473999,
+      "learning_rate": 8.399880185025438e-06,
+      "loss": 0.2056,
+      "step": 1727
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.4849991798400879,
+      "learning_rate": 8.357337165935675e-06,
+      "loss": 0.1321,
+      "step": 1728
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.4412074089050293,
+      "learning_rate": 8.314892327719936e-06,
+      "loss": 0.1139,
+      "step": 1729
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.4538963735103607,
+      "learning_rate": 8.27254577045039e-06,
+      "loss": 0.1128,
+      "step": 1730
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5626980066299438,
+      "learning_rate": 8.230297593967461e-06,
+      "loss": 0.1998,
+      "step": 1731
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5795850157737732,
+      "learning_rate": 8.188147897879667e-06,
+      "loss": 0.1888,
+      "step": 1732
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5094465017318726,
+      "learning_rate": 8.146096781563284e-06,
+      "loss": 0.1419,
+      "step": 1733
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.48453688621520996,
+      "learning_rate": 8.104144344162229e-06,
+      "loss": 0.1707,
+      "step": 1734
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5459434390068054,
+      "learning_rate": 8.062290684587698e-06,
+      "loss": 0.167,
+      "step": 1735
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.6594077944755554,
+      "learning_rate": 8.02053590151805e-06,
+      "loss": 0.2591,
+      "step": 1736
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5281089544296265,
+      "learning_rate": 7.978880093398516e-06,
+      "loss": 0.1698,
+      "step": 1737
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.64583820104599,
+      "learning_rate": 7.937323358440935e-06,
+      "loss": 0.2043,
+      "step": 1738
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.6523512601852417,
+      "learning_rate": 7.89586579462362e-06,
+      "loss": 0.2278,
+      "step": 1739
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.42586687207221985,
+      "learning_rate": 7.854507499691005e-06,
+      "loss": 0.1319,
+      "step": 1740
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.46904200315475464,
+      "learning_rate": 7.813248571153542e-06,
+      "loss": 0.144,
+      "step": 1741
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5434065461158752,
+      "learning_rate": 7.772089106287344e-06,
+      "loss": 0.1936,
+      "step": 1742
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.4898480176925659,
+      "learning_rate": 7.731029202134078e-06,
+      "loss": 0.1783,
+      "step": 1743
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5367664694786072,
+      "learning_rate": 7.690068955500624e-06,
+      "loss": 0.1678,
+      "step": 1744
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.5709487795829773,
+      "learning_rate": 7.649208462958935e-06,
+      "loss": 0.189,
+      "step": 1745
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.6088728904724121,
+      "learning_rate": 7.608447820845771e-06,
+      "loss": 0.2121,
+      "step": 1746
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.49801066517829895,
+      "learning_rate": 7.5677871252624485e-06,
+      "loss": 0.1491,
+      "step": 1747
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.4537121057510376,
+      "learning_rate": 7.527226472074678e-06,
+      "loss": 0.1183,
+      "step": 1748
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5984990000724792,
+      "learning_rate": 7.486765956912262e-06,
+      "loss": 0.1941,
+      "step": 1749
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.7296456694602966,
+      "learning_rate": 7.446405675168938e-06,
+      "loss": 0.2947,
+      "step": 1750
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.8127968311309814,
+      "learning_rate": 7.4061457220021e-06,
+      "loss": 0.228,
+      "step": 1751
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.6742526888847351,
+      "learning_rate": 7.365986192332625e-06,
+      "loss": 0.2431,
+      "step": 1752
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5769400000572205,
+      "learning_rate": 7.325927180844589e-06,
+      "loss": 0.2149,
+      "step": 1753
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5331418514251709,
+      "learning_rate": 7.285968781985092e-06,
+      "loss": 0.1694,
+      "step": 1754
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.632719874382019,
+      "learning_rate": 7.246111089964042e-06,
+      "loss": 0.2146,
+      "step": 1755
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.6811960935592651,
+      "learning_rate": 7.206354198753862e-06,
+      "loss": 0.2244,
+      "step": 1756
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5719199776649475,
+      "learning_rate": 7.166698202089367e-06,
+      "loss": 0.1971,
+      "step": 1757
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.700746476650238,
+      "learning_rate": 7.127143193467445e-06,
+      "loss": 0.2231,
+      "step": 1758
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.7047358155250549,
+      "learning_rate": 7.087689266146935e-06,
+      "loss": 0.2302,
+      "step": 1759
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.4994795322418213,
+      "learning_rate": 7.048336513148307e-06,
+      "loss": 0.1588,
+      "step": 1760
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.40862783789634705,
+      "learning_rate": 7.009085027253543e-06,
+      "loss": 0.1175,
+      "step": 1761
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.8352911472320557,
+      "learning_rate": 6.969934901005809e-06,
+      "loss": 0.2179,
+      "step": 1762
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.4421202838420868,
+      "learning_rate": 6.930886226709343e-06,
+      "loss": 0.1018,
+      "step": 1763
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5586758852005005,
+      "learning_rate": 6.8919390964291895e-06,
+      "loss": 0.1753,
+      "step": 1764
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5938215255737305,
+      "learning_rate": 6.853093601990945e-06,
+      "loss": 0.2415,
+      "step": 1765
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.6451236605644226,
+      "learning_rate": 6.814349834980621e-06,
+      "loss": 0.2309,
+      "step": 1766
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5209770202636719,
+      "learning_rate": 6.775707886744342e-06,
+      "loss": 0.1677,
+      "step": 1767
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.554765522480011,
+      "learning_rate": 6.7371678483882264e-06,
+      "loss": 0.1955,
+      "step": 1768
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5225728154182434,
+      "learning_rate": 6.698729810778065e-06,
+      "loss": 0.1696,
+      "step": 1769
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5427529215812683,
+      "learning_rate": 6.660393864539222e-06,
+      "loss": 0.1587,
+      "step": 1770
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5016469359397888,
+      "learning_rate": 6.622160100056296e-06,
+      "loss": 0.1434,
+      "step": 1771
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.42081448435783386,
+      "learning_rate": 6.58402860747302e-06,
+      "loss": 0.1258,
+      "step": 1772
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.41252243518829346,
+      "learning_rate": 6.5459994766919945e-06,
+      "loss": 0.1287,
+      "step": 1773
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5363560318946838,
+      "learning_rate": 6.508072797374454e-06,
+      "loss": 0.1801,
+      "step": 1774
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.4970044195652008,
+      "learning_rate": 6.470248658940115e-06,
+      "loss": 0.1447,
+      "step": 1775
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5494373440742493,
+      "learning_rate": 6.4325271505669025e-06,
+      "loss": 0.1573,
+      "step": 1776
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.6385138630867004,
+      "learning_rate": 6.394908361190805e-06,
+      "loss": 0.2057,
+      "step": 1777
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.668546199798584,
+      "learning_rate": 6.3573923795056e-06,
+      "loss": 0.1765,
+      "step": 1778
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5419114828109741,
+      "learning_rate": 6.319979293962691e-06,
+      "loss": 0.2021,
+      "step": 1779
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5417685508728027,
+      "learning_rate": 6.282669192770896e-06,
+      "loss": 0.1554,
+      "step": 1780
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.6172122955322266,
+      "learning_rate": 6.245462163896188e-06,
+      "loss": 0.1656,
+      "step": 1781
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5292225480079651,
+      "learning_rate": 6.208358295061572e-06,
+      "loss": 0.1814,
+      "step": 1782
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.655319094657898,
+      "learning_rate": 6.171357673746797e-06,
+      "loss": 0.2053,
+      "step": 1783
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.605711817741394,
+      "learning_rate": 6.134460387188206e-06,
+      "loss": 0.16,
+      "step": 1784
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5367722511291504,
+      "learning_rate": 6.097666522378498e-06,
+      "loss": 0.1793,
+      "step": 1785
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5480538010597229,
+      "learning_rate": 6.060976166066546e-06,
+      "loss": 0.1523,
+      "step": 1786
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.6719861626625061,
+      "learning_rate": 6.024389404757164e-06,
+      "loss": 0.1799,
+      "step": 1787
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5512567162513733,
+      "learning_rate": 5.9879063247109405e-06,
+      "loss": 0.147,
+      "step": 1788
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5770524144172668,
+      "learning_rate": 5.951527011944008e-06,
+      "loss": 0.1995,
+      "step": 1789
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6128679513931274,
+      "learning_rate": 5.915251552227829e-06,
+      "loss": 0.2136,
+      "step": 1790
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.615993082523346,
+      "learning_rate": 5.879080031089046e-06,
+      "loss": 0.2247,
+      "step": 1791
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5884333252906799,
+      "learning_rate": 5.8430125338092115e-06,
+      "loss": 0.1833,
+      "step": 1792
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5691912174224854,
+      "learning_rate": 5.807049145424648e-06,
+      "loss": 0.1714,
+      "step": 1793
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5145797729492188,
+      "learning_rate": 5.771189950726191e-06,
+      "loss": 0.1593,
+      "step": 1794
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5392258167266846,
+      "learning_rate": 5.7354350342590566e-06,
+      "loss": 0.1624,
+      "step": 1795
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.43291908502578735,
+      "learning_rate": 5.699784480322568e-06,
+      "loss": 0.1309,
+      "step": 1796
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6070436835289001,
+      "learning_rate": 5.664238372970015e-06,
+      "loss": 0.1606,
+      "step": 1797
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5545229315757751,
+      "learning_rate": 5.628796796008434e-06,
+      "loss": 0.1901,
+      "step": 1798
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5199242234230042,
+      "learning_rate": 5.5934598329983874e-06,
+      "loss": 0.1515,
+      "step": 1799
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.7784616947174072,
+      "learning_rate": 5.558227567253832e-06,
+      "loss": 0.2903,
+      "step": 1800
+    },
+    {
+      "epoch": 1.71,
+      "eval_loss": 0.29014164209365845,
+      "eval_runtime": 360.7417,
+      "eval_samples_per_second": 0.945,
+      "eval_steps_per_second": 0.238,
+      "step": 1800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.7077049016952515,
+      "learning_rate": 5.52310008184182e-06,
+      "loss": 0.2263,
+      "step": 1801
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5775318741798401,
+      "learning_rate": 5.488077459582425e-06,
+      "loss": 0.1389,
+      "step": 1802
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5910678505897522,
+      "learning_rate": 5.453159783048433e-06,
+      "loss": 0.146,
+      "step": 1803
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5038835406303406,
+      "learning_rate": 5.418347134565249e-06,
+      "loss": 0.1456,
+      "step": 1804
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5334418416023254,
+      "learning_rate": 5.383639596210604e-06,
+      "loss": 0.2042,
+      "step": 1805
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5270768404006958,
+      "learning_rate": 5.349037249814442e-06,
+      "loss": 0.1486,
+      "step": 1806
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.48578596115112305,
+      "learning_rate": 5.314540176958699e-06,
+      "loss": 0.1161,
+      "step": 1807
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5829055905342102,
+      "learning_rate": 5.28014845897708e-06,
+      "loss": 0.2223,
+      "step": 1808
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5074483156204224,
+      "learning_rate": 5.24586217695493e-06,
+      "loss": 0.1554,
+      "step": 1809
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.5854430198669434,
+      "learning_rate": 5.211681411728969e-06,
+      "loss": 0.2032,
+      "step": 1810
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.520721971988678,
+      "learning_rate": 5.177606243887184e-06,
+      "loss": 0.1781,
+      "step": 1811
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.3583303391933441,
+      "learning_rate": 5.143636753768549e-06,
+      "loss": 0.1103,
+      "step": 1812
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4149814546108246,
+      "learning_rate": 5.109773021462921e-06,
+      "loss": 0.1114,
+      "step": 1813
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.523317813873291,
+      "learning_rate": 5.076015126810784e-06,
+      "loss": 0.1821,
+      "step": 1814
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.5302857160568237,
+      "learning_rate": 5.042363149403107e-06,
+      "loss": 0.201,
+      "step": 1815
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.6901510953903198,
+      "learning_rate": 5.008817168581137e-06,
+      "loss": 0.2446,
+      "step": 1816
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.6927199959754944,
+      "learning_rate": 4.975377263436193e-06,
+      "loss": 0.2071,
+      "step": 1817
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.6235139966011047,
+      "learning_rate": 4.94204351280953e-06,
+      "loss": 0.253,
+      "step": 1818
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.5554899573326111,
+      "learning_rate": 4.908815995292082e-06,
+      "loss": 0.2185,
+      "step": 1819
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4405827820301056,
+      "learning_rate": 4.8756947892243725e-06,
+      "loss": 0.1634,
+      "step": 1820
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5380933284759521,
+      "learning_rate": 4.842679972696213e-06,
+      "loss": 0.1506,
+      "step": 1821
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5965523719787598,
+      "learning_rate": 4.809771623546627e-06,
+      "loss": 0.1982,
+      "step": 1822
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5557235479354858,
+      "learning_rate": 4.7769698193636135e-06,
+      "loss": 0.1655,
+      "step": 1823
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5408238768577576,
+      "learning_rate": 4.744274637483936e-06,
+      "loss": 0.154,
+      "step": 1824
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.4990658760070801,
+      "learning_rate": 4.711686154993028e-06,
+      "loss": 0.1541,
+      "step": 1825
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.6194915771484375,
+      "learning_rate": 4.6792044487247e-06,
+      "loss": 0.2069,
+      "step": 1826
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5993006825447083,
+      "learning_rate": 4.646829595261071e-06,
+      "loss": 0.2059,
+      "step": 1827
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5607315301895142,
+      "learning_rate": 4.614561670932288e-06,
+      "loss": 0.1504,
+      "step": 1828
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5842840671539307,
+      "learning_rate": 4.582400751816435e-06,
+      "loss": 0.1816,
+      "step": 1829
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5975632071495056,
+      "learning_rate": 4.550346913739256e-06,
+      "loss": 0.1796,
+      "step": 1830
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.688764750957489,
+      "learning_rate": 4.5184002322740785e-06,
+      "loss": 0.2402,
+      "step": 1831
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.6619789600372314,
+      "learning_rate": 4.486560782741578e-06,
+      "loss": 0.2536,
+      "step": 1832
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.6045652031898499,
+      "learning_rate": 4.454828640209574e-06,
+      "loss": 0.2001,
+      "step": 1833
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.61651211977005,
+      "learning_rate": 4.423203879492943e-06,
+      "loss": 0.193,
+      "step": 1834
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.5226234793663025,
+      "learning_rate": 4.391686575153331e-06,
+      "loss": 0.1616,
+      "step": 1835
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.5356714725494385,
+      "learning_rate": 4.360276801499086e-06,
+      "loss": 0.1791,
+      "step": 1836
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.6690487265586853,
+      "learning_rate": 4.328974632584992e-06,
+      "loss": 0.2162,
+      "step": 1837
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.5904943943023682,
+      "learning_rate": 4.29778014221216e-06,
+      "loss": 0.1931,
+      "step": 1838
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.5628692507743835,
+      "learning_rate": 4.266693403927802e-06,
+      "loss": 0.1777,
+      "step": 1839
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.49221864342689514,
+      "learning_rate": 4.2357144910251e-06,
+      "loss": 0.1435,
+      "step": 1840
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.4652506709098816,
+      "learning_rate": 4.20484347654303e-06,
+      "loss": 0.146,
+      "step": 1841
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.6496181488037109,
+      "learning_rate": 4.174080433266136e-06,
+      "loss": 0.1834,
+      "step": 1842
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.6121329665184021,
+      "learning_rate": 4.1434254337244404e-06,
+      "loss": 0.2422,
+      "step": 1843
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5939140915870667,
+      "learning_rate": 4.112878550193194e-06,
+      "loss": 0.1691,
+      "step": 1844
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5642305612564087,
+      "learning_rate": 4.082439854692782e-06,
+      "loss": 0.1821,
+      "step": 1845
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.7495043873786926,
+      "learning_rate": 4.05210941898847e-06,
+      "loss": 0.2244,
+      "step": 1846
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.6956668496131897,
+      "learning_rate": 4.021887314590323e-06,
+      "loss": 0.2272,
+      "step": 1847
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.6110973358154297,
+      "learning_rate": 3.991773612752952e-06,
+      "loss": 0.2428,
+      "step": 1848
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5061264038085938,
+      "learning_rate": 3.961768384475428e-06,
+      "loss": 0.1356,
+      "step": 1849
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5810338258743286,
+      "learning_rate": 3.9318717005010495e-06,
+      "loss": 0.1876,
+      "step": 1850
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.6085287928581238,
+      "learning_rate": 3.902083631317193e-06,
+      "loss": 0.1965,
+      "step": 1851
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5208376049995422,
+      "learning_rate": 3.872404247155193e-06,
+      "loss": 0.141,
+      "step": 1852
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5519346594810486,
+      "learning_rate": 3.842833617990077e-06,
+      "loss": 0.1815,
+      "step": 1853
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.586296021938324,
+      "learning_rate": 3.8133718135405285e-06,
+      "loss": 0.1955,
+      "step": 1854
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.46704259514808655,
+      "learning_rate": 3.784018903268588e-06,
+      "loss": 0.1424,
+      "step": 1855
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5107570290565491,
+      "learning_rate": 3.754774956379614e-06,
+      "loss": 0.1565,
+      "step": 1856
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.559743344783783,
+      "learning_rate": 3.7256400418220262e-06,
+      "loss": 0.1525,
+      "step": 1857
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5526383519172668,
+      "learning_rate": 3.696614228287187e-06,
+      "loss": 0.2024,
+      "step": 1858
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.6702591776847839,
+      "learning_rate": 3.6676975842092508e-06,
+      "loss": 0.2615,
+      "step": 1859
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5086106061935425,
+      "learning_rate": 3.6388901777649477e-06,
+      "loss": 0.1493,
+      "step": 1860
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.7395433187484741,
+      "learning_rate": 3.6101920768734975e-06,
+      "loss": 0.1989,
+      "step": 1861
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.6108511686325073,
+      "learning_rate": 3.581603349196372e-06,
+      "loss": 0.2126,
+      "step": 1862
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.6634174585342407,
+      "learning_rate": 3.553124062137203e-06,
+      "loss": 0.1792,
+      "step": 1863
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5514046549797058,
+      "learning_rate": 3.5247542828415747e-06,
+      "loss": 0.1688,
+      "step": 1864
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5157788395881653,
+      "learning_rate": 3.4964940781969104e-06,
+      "loss": 0.1674,
+      "step": 1865
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5924208164215088,
+      "learning_rate": 3.468343514832251e-06,
+      "loss": 0.1727,
+      "step": 1866
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5168029069900513,
+      "learning_rate": 3.440302659118172e-06,
+      "loss": 0.154,
+      "step": 1867
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5888129472732544,
+      "learning_rate": 3.4123715771665786e-06,
+      "loss": 0.1296,
+      "step": 1868
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.6567624807357788,
+      "learning_rate": 3.384550334830555e-06,
+      "loss": 0.2119,
+      "step": 1869
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.6347790360450745,
+      "learning_rate": 3.356838997704226e-06,
+      "loss": 0.1986,
+      "step": 1870
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.541713535785675,
+      "learning_rate": 3.3292376311225835e-06,
+      "loss": 0.138,
+      "step": 1871
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.5737549066543579,
+      "learning_rate": 3.3017463001613626e-06,
+      "loss": 0.2148,
+      "step": 1872
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.512398898601532,
+      "learning_rate": 3.274365069636831e-06,
+      "loss": 0.1492,
+      "step": 1873
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.48606324195861816,
+      "learning_rate": 3.2470940041057106e-06,
+      "loss": 0.173,
+      "step": 1874
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.6552368402481079,
+      "learning_rate": 3.2199331678649803e-06,
+      "loss": 0.1732,
+      "step": 1875
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5576645135879517,
+      "learning_rate": 3.1928826249516987e-06,
+      "loss": 0.1926,
+      "step": 1876
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.592711865901947,
+      "learning_rate": 3.165942439142927e-06,
+      "loss": 0.1879,
+      "step": 1877
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.4957755208015442,
+      "learning_rate": 3.139112673955513e-06,
+      "loss": 0.1647,
+      "step": 1878
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5901917815208435,
+      "learning_rate": 3.112393392645985e-06,
+      "loss": 0.1835,
+      "step": 1879
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.6506839990615845,
+      "learning_rate": 3.08578465821035e-06,
+      "loss": 0.275,
+      "step": 1880
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.6301924586296082,
+      "learning_rate": 3.0592865333840204e-06,
+      "loss": 0.2242,
+      "step": 1881
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.6617907285690308,
+      "learning_rate": 3.0328990806415934e-06,
+      "loss": 0.1901,
+      "step": 1882
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5383888483047485,
+      "learning_rate": 3.006622362196748e-06,
+      "loss": 0.1751,
+      "step": 1883
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.6773173809051514,
+      "learning_rate": 2.9804564400020994e-06,
+      "loss": 0.2043,
+      "step": 1884
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5650835633277893,
+      "learning_rate": 2.954401375748994e-06,
+      "loss": 0.1859,
+      "step": 1885
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5167426466941833,
+      "learning_rate": 2.9284572308674596e-06,
+      "loss": 0.1738,
+      "step": 1886
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5418599247932434,
+      "learning_rate": 2.9026240665259717e-06,
+      "loss": 0.1784,
+      "step": 1887
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.579948365688324,
+      "learning_rate": 2.8769019436313715e-06,
+      "loss": 0.2147,
+      "step": 1888
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5538483262062073,
+      "learning_rate": 2.851290922828681e-06,
+      "loss": 0.1446,
+      "step": 1889
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5067595839500427,
+      "learning_rate": 2.825791064500993e-06,
+      "loss": 0.1666,
+      "step": 1890
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5989078283309937,
+      "learning_rate": 2.800402428769294e-06,
+      "loss": 0.2095,
+      "step": 1891
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.47605156898498535,
+      "learning_rate": 2.7751250754923573e-06,
+      "loss": 0.1428,
+      "step": 1892
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.49020832777023315,
+      "learning_rate": 2.7499590642665774e-06,
+      "loss": 0.1276,
+      "step": 1893
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.6077764630317688,
+      "learning_rate": 2.7249044544258363e-06,
+      "loss": 0.1951,
+      "step": 1894
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5305209755897522,
+      "learning_rate": 2.699961305041382e-06,
+      "loss": 0.1432,
+      "step": 1895
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5732347965240479,
+      "learning_rate": 2.675129674921639e-06,
+      "loss": 0.2081,
+      "step": 1896
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.4637000858783722,
+      "learning_rate": 2.6504096226121376e-06,
+      "loss": 0.1503,
+      "step": 1897
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.588767409324646,
+      "learning_rate": 2.625801206395312e-06,
+      "loss": 0.1979,
+      "step": 1898
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.644022524356842,
+      "learning_rate": 2.601304484290423e-06,
+      "loss": 0.2002,
+      "step": 1899
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.6414205431938171,
+      "learning_rate": 2.576919514053355e-06,
+      "loss": 0.2506,
+      "step": 1900
+    },
+    {
+      "epoch": 1.8,
+      "eval_loss": 0.2901623547077179,
+      "eval_runtime": 376.7208,
+      "eval_samples_per_second": 0.905,
+      "eval_steps_per_second": 0.228,
+      "step": 1900
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.6934900283813477,
+      "learning_rate": 2.5526463531765465e-06,
+      "loss": 0.2258,
+      "step": 1901
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.6121662855148315,
+      "learning_rate": 2.528485058888813e-06,
+      "loss": 0.1935,
+      "step": 1902
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.7360637187957764,
+      "learning_rate": 2.5044356881552044e-06,
+      "loss": 0.2149,
+      "step": 1903
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.7367038130760193,
+      "learning_rate": 2.48049829767692e-06,
+      "loss": 0.2557,
+      "step": 1904
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.6138616800308228,
+      "learning_rate": 2.456672943891114e-06,
+      "loss": 0.2147,
+      "step": 1905
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.42625892162323,
+      "learning_rate": 2.4329596829708144e-06,
+      "loss": 0.1285,
+      "step": 1906
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.6234516501426697,
+      "learning_rate": 2.4093585708247492e-06,
+      "loss": 0.2156,
+      "step": 1907
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.5818755626678467,
+      "learning_rate": 2.385869663097251e-06,
+      "loss": 0.1994,
+      "step": 1908
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.6006777286529541,
+      "learning_rate": 2.3624930151680878e-06,
+      "loss": 0.2469,
+      "step": 1909
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.4501885175704956,
+      "learning_rate": 2.3392286821523724e-06,
+      "loss": 0.1186,
+      "step": 1910
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.5956398844718933,
+      "learning_rate": 2.316076718900412e-06,
+      "loss": 0.2159,
+      "step": 1911
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.6961279511451721,
+      "learning_rate": 2.2930371799975594e-06,
+      "loss": 0.278,
+      "step": 1912
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.5073667764663696,
+      "learning_rate": 2.2701101197641237e-06,
+      "loss": 0.0936,
+      "step": 1913
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.8118249773979187,
+      "learning_rate": 2.2472955922552163e-06,
+      "loss": 0.1759,
+      "step": 1914
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.6911466717720032,
+      "learning_rate": 2.2245936512606314e-06,
+      "loss": 0.1994,
+      "step": 1915
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.7075309753417969,
+      "learning_rate": 2.202004350304715e-06,
+      "loss": 0.27,
+      "step": 1916
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.6095204949378967,
+      "learning_rate": 2.179527742646248e-06,
+      "loss": 0.1509,
+      "step": 1917
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5882623195648193,
+      "learning_rate": 2.157163881278312e-06,
+      "loss": 0.1785,
+      "step": 1918
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5185151100158691,
+      "learning_rate": 2.1349128189281586e-06,
+      "loss": 0.1595,
+      "step": 1919
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5481693148612976,
+      "learning_rate": 2.112774608057111e-06,
+      "loss": 0.1638,
+      "step": 1920
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5209944844245911,
+      "learning_rate": 2.0907493008604007e-06,
+      "loss": 0.1783,
+      "step": 1921
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.43272966146469116,
+      "learning_rate": 2.068836949267089e-06,
+      "loss": 0.129,
+      "step": 1922
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.42314204573631287,
+      "learning_rate": 2.047037604939894e-06,
+      "loss": 0.1055,
+      "step": 1923
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.6561240553855896,
+      "learning_rate": 2.0253513192751373e-06,
+      "loss": 0.1929,
+      "step": 1924
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5800967216491699,
+      "learning_rate": 2.003778143402535e-06,
+      "loss": 0.1599,
+      "step": 1925
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.40080785751342773,
+      "learning_rate": 1.982318128185151e-06,
+      "loss": 0.0993,
+      "step": 1926
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5818573832511902,
+      "learning_rate": 1.960971324219263e-06,
+      "loss": 0.1966,
+      "step": 1927
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5501309037208557,
+      "learning_rate": 1.9397377818341944e-06,
+      "loss": 0.1803,
+      "step": 1928
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5448240041732788,
+      "learning_rate": 1.9186175510922664e-06,
+      "loss": 0.1631,
+      "step": 1929
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5137999057769775,
+      "learning_rate": 1.8976106817886196e-06,
+      "loss": 0.1659,
+      "step": 1930
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6308398246765137,
+      "learning_rate": 1.876717223451141e-06,
+      "loss": 0.1998,
+      "step": 1931
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6079750061035156,
+      "learning_rate": 1.855937225340315e-06,
+      "loss": 0.1972,
+      "step": 1932
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5553622245788574,
+      "learning_rate": 1.835270736449135e-06,
+      "loss": 0.155,
+      "step": 1933
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6064857244491577,
+      "learning_rate": 1.8147178055029579e-06,
+      "loss": 0.1827,
+      "step": 1934
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5615193247795105,
+      "learning_rate": 1.794278480959416e-06,
+      "loss": 0.1673,
+      "step": 1935
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5754190683364868,
+      "learning_rate": 1.7739528110083004e-06,
+      "loss": 0.1569,
+      "step": 1936
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.6011772751808167,
+      "learning_rate": 1.7537408435714053e-06,
+      "loss": 0.1613,
+      "step": 1937
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.5278980135917664,
+      "learning_rate": 1.7336426263024897e-06,
+      "loss": 0.159,
+      "step": 1938
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.6010107398033142,
+      "learning_rate": 1.7136582065870877e-06,
+      "loss": 0.1872,
+      "step": 1939
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.686768651008606,
+      "learning_rate": 1.6937876315424705e-06,
+      "loss": 0.2399,
+      "step": 1940
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.7978424429893494,
+      "learning_rate": 1.674030948017463e-06,
+      "loss": 0.2631,
+      "step": 1941
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.5631486177444458,
+      "learning_rate": 1.6543882025923886e-06,
+      "loss": 0.1622,
+      "step": 1942
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.42832475900650024,
+      "learning_rate": 1.6348594415789288e-06,
+      "loss": 0.1542,
+      "step": 1943
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.7677817344665527,
+      "learning_rate": 1.6154447110200255e-06,
+      "loss": 0.2232,
+      "step": 1944
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.50302654504776,
+      "learning_rate": 1.5961440566897912e-06,
+      "loss": 0.1462,
+      "step": 1945
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.5557914972305298,
+      "learning_rate": 1.5769575240933422e-06,
+      "loss": 0.146,
+      "step": 1946
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.5259419679641724,
+      "learning_rate": 1.5578851584667652e-06,
+      "loss": 0.1473,
+      "step": 1947
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5671291351318359,
+      "learning_rate": 1.5389270047769578e-06,
+      "loss": 0.1713,
+      "step": 1948
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.46985188126564026,
+      "learning_rate": 1.5200831077215428e-06,
+      "loss": 0.125,
+      "step": 1949
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5260842442512512,
+      "learning_rate": 1.5013535117287647e-06,
+      "loss": 0.1764,
+      "step": 1950
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5348197817802429,
+      "learning_rate": 1.4827382609573781e-06,
+      "loss": 0.133,
+      "step": 1951
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5998916625976562,
+      "learning_rate": 1.4642373992965364e-06,
+      "loss": 0.2082,
+      "step": 1952
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.521109938621521,
+      "learning_rate": 1.4458509703657196e-06,
+      "loss": 0.1963,
+      "step": 1953
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5232467651367188,
+      "learning_rate": 1.4275790175145908e-06,
+      "loss": 0.1342,
+      "step": 1954
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5119155645370483,
+      "learning_rate": 1.4094215838229176e-06,
+      "loss": 0.1468,
+      "step": 1955
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6162364482879639,
+      "learning_rate": 1.3913787121004719e-06,
+      "loss": 0.2073,
+      "step": 1956
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6221475601196289,
+      "learning_rate": 1.3734504448869145e-06,
+      "loss": 0.1372,
+      "step": 1957
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5454715490341187,
+      "learning_rate": 1.3556368244517114e-06,
+      "loss": 0.2036,
+      "step": 1958
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.637290894985199,
+      "learning_rate": 1.3379378927940167e-06,
+      "loss": 0.259,
+      "step": 1959
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5475330948829651,
+      "learning_rate": 1.3203536916425841e-06,
+      "loss": 0.1199,
+      "step": 1960
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.7231860160827637,
+      "learning_rate": 1.3028842624556893e-06,
+      "loss": 0.2889,
+      "step": 1961
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.612337052822113,
+      "learning_rate": 1.2855296464209688e-06,
+      "loss": 0.169,
+      "step": 1962
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5270692706108093,
+      "learning_rate": 1.268289884455409e-06,
+      "loss": 0.1665,
+      "step": 1963
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.4391113221645355,
+      "learning_rate": 1.2511650172051637e-06,
+      "loss": 0.1182,
+      "step": 1964
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.5525397658348083,
+      "learning_rate": 1.2341550850455352e-06,
+      "loss": 0.2095,
+      "step": 1965
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6357365846633911,
+      "learning_rate": 1.217260128080816e-06,
+      "loss": 0.2119,
+      "step": 1966
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.4750954210758209,
+      "learning_rate": 1.2004801861442371e-06,
+      "loss": 0.127,
+      "step": 1967
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.7122442722320557,
+      "learning_rate": 1.1838152987978578e-06,
+      "loss": 0.2634,
+      "step": 1968
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3684752881526947,
+      "learning_rate": 1.1672655053324655e-06,
+      "loss": 0.1047,
+      "step": 1969
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.4597022235393524,
+      "learning_rate": 1.1508308447674975e-06,
+      "loss": 0.1087,
+      "step": 1970
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5127688050270081,
+      "learning_rate": 1.1345113558509423e-06,
+      "loss": 0.1343,
+      "step": 1971
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.6507446765899658,
+      "learning_rate": 1.1183070770592441e-06,
+      "loss": 0.1748,
+      "step": 1972
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.512505292892456,
+      "learning_rate": 1.1022180465972199e-06,
+      "loss": 0.1432,
+      "step": 1973
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5194410085678101,
+      "learning_rate": 1.0862443023979651e-06,
+      "loss": 0.1772,
+      "step": 1974
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.6167486310005188,
+      "learning_rate": 1.0703858821227541e-06,
+      "loss": 0.1961,
+      "step": 1975
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5342720746994019,
+      "learning_rate": 1.0546428231609895e-06,
+      "loss": 0.1198,
+      "step": 1976
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5644630789756775,
+      "learning_rate": 1.0390151626300526e-06,
+      "loss": 0.1488,
+      "step": 1977
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.6594527959823608,
+      "learning_rate": 1.0235029373752758e-06,
+      "loss": 0.2731,
+      "step": 1978
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.4499061107635498,
+      "learning_rate": 1.0081061839698258e-06,
+      "loss": 0.1257,
+      "step": 1979
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5581822991371155,
+      "learning_rate": 9.928249387145982e-07,
+      "loss": 0.1409,
+      "step": 1980
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5295170545578003,
+      "learning_rate": 9.776592376381955e-07,
+      "loss": 0.1747,
+      "step": 1981
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.6919394135475159,
+      "learning_rate": 9.6260911649676e-07,
+      "loss": 0.1795,
+      "step": 1982
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.6031213402748108,
+      "learning_rate": 9.476746107739576e-07,
+      "loss": 0.192,
+      "step": 1983
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5239519476890564,
+      "learning_rate": 9.32855755680867e-07,
+      "loss": 0.1833,
+      "step": 1984
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5798040628433228,
+      "learning_rate": 9.181525861558849e-07,
+      "loss": 0.1889,
+      "step": 1985
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5663641095161438,
+      "learning_rate": 9.035651368646648e-07,
+      "loss": 0.183,
+      "step": 1986
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.6420210599899292,
+      "learning_rate": 8.89093442200023e-07,
+      "loss": 0.2077,
+      "step": 1987
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5905512571334839,
+      "learning_rate": 8.747375362818666e-07,
+      "loss": 0.1858,
+      "step": 1988
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.48887673020362854,
+      "learning_rate": 8.604974529571042e-07,
+      "loss": 0.1234,
+      "step": 1989
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.5324652791023254,
+      "learning_rate": 8.463732257995571e-07,
+      "loss": 0.1588,
+      "step": 1990
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.6185021996498108,
+      "learning_rate": 8.323648881099211e-07,
+      "loss": 0.1816,
+      "step": 1991
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.6033267378807068,
+      "learning_rate": 8.184724729156379e-07,
+      "loss": 0.1799,
+      "step": 1992
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.5450901985168457,
+      "learning_rate": 8.046960129708348e-07,
+      "loss": 0.1317,
+      "step": 1993
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4834456741809845,
+      "learning_rate": 7.910355407562742e-07,
+      "loss": 0.1497,
+      "step": 1994
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.6220552325248718,
+      "learning_rate": 7.774910884792319e-07,
+      "loss": 0.17,
+      "step": 1995
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.5117090940475464,
+      "learning_rate": 7.640626880734581e-07,
+      "loss": 0.1607,
+      "step": 1996
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.5739068984985352,
+      "learning_rate": 7.507503711990771e-07,
+      "loss": 0.2166,
+      "step": 1997
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.5173081755638123,
+      "learning_rate": 7.375541692425326e-07,
+      "loss": 0.1973,
+      "step": 1998
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.5910633206367493,
+      "learning_rate": 7.244741133164978e-07,
+      "loss": 0.2178,
+      "step": 1999
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.4375127851963043,
+      "learning_rate": 7.115102342598101e-07,
+      "loss": 0.1516,
+      "step": 2000
+    },
+    {
+      "epoch": 1.9,
+      "eval_loss": 0.28917258977890015,
+      "eval_runtime": 380.1345,
+      "eval_samples_per_second": 0.897,
+      "eval_steps_per_second": 0.226,
+      "step": 2000
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5685524344444275,
+      "learning_rate": 6.986625626373977e-07,
+      "loss": 0.1853,
+      "step": 2001
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.606907844543457,
+      "learning_rate": 6.859311287402081e-07,
+      "loss": 0.1655,
+      "step": 2002
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.49601927399635315,
+      "learning_rate": 6.733159625851304e-07,
+      "loss": 0.1595,
+      "step": 2003
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.8226091265678406,
+      "learning_rate": 6.608170939149283e-07,
+      "loss": 0.3011,
+      "step": 2004
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5906044244766235,
+      "learning_rate": 6.484345521981849e-07,
+      "loss": 0.2237,
+      "step": 2005
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5248388051986694,
+      "learning_rate": 6.361683666291973e-07,
+      "loss": 0.131,
+      "step": 2006
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.4451999366283417,
+      "learning_rate": 6.24018566127954e-07,
+      "loss": 0.1438,
+      "step": 2007
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5558008551597595,
+      "learning_rate": 6.119851793400189e-07,
+      "loss": 0.1647,
+      "step": 2008
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.6137389540672302,
+      "learning_rate": 6.000682346365083e-07,
+      "loss": 0.162,
+      "step": 2009
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5638387799263,
+      "learning_rate": 5.882677601139919e-07,
+      "loss": 0.1814,
+      "step": 2010
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6443926692008972,
+      "learning_rate": 5.765837835944309e-07,
+      "loss": 0.2643,
+      "step": 2011
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6227025389671326,
+      "learning_rate": 5.650163326251345e-07,
+      "loss": 0.1894,
+      "step": 2012
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5981712341308594,
+      "learning_rate": 5.535654344786756e-07,
+      "loss": 0.185,
+      "step": 2013
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6320492625236511,
+      "learning_rate": 5.422311161528193e-07,
+      "loss": 0.1747,
+      "step": 2014
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6899151802062988,
+      "learning_rate": 5.310134043704896e-07,
+      "loss": 0.2454,
+      "step": 2015
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5967755913734436,
+      "learning_rate": 5.199123255796634e-07,
+      "loss": 0.1828,
+      "step": 2016
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5143026113510132,
+      "learning_rate": 5.089279059533658e-07,
+      "loss": 0.1531,
+      "step": 2017
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6956952810287476,
+      "learning_rate": 4.980601713895305e-07,
+      "loss": 0.2294,
+      "step": 2018
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6141075491905212,
+      "learning_rate": 4.873091475110281e-07,
+      "loss": 0.1814,
+      "step": 2019
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.532913327217102,
+      "learning_rate": 4.7667485966552685e-07,
+      "loss": 0.1785,
+      "step": 2020
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.655834436416626,
+      "learning_rate": 4.66157332925482e-07,
+      "loss": 0.15,
+      "step": 2021
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.4919854700565338,
+      "learning_rate": 4.55756592088058e-07,
+      "loss": 0.1685,
+      "step": 2022
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5122151970863342,
+      "learning_rate": 4.454726616750726e-07,
+      "loss": 0.1559,
+      "step": 2023
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5295954346656799,
+      "learning_rate": 4.353055659329419e-07,
+      "loss": 0.1568,
+      "step": 2024
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6130471229553223,
+      "learning_rate": 4.2525532883261886e-07,
+      "loss": 0.1926,
+      "step": 2025
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.45136332511901855,
+      "learning_rate": 4.153219740695435e-07,
+      "loss": 0.1572,
+      "step": 2026
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.7034523487091064,
+      "learning_rate": 4.0550552506357644e-07,
+      "loss": 0.2333,
+      "step": 2027
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6193264126777649,
+      "learning_rate": 3.958060049589485e-07,
+      "loss": 0.1998,
+      "step": 2028
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5168358087539673,
+      "learning_rate": 3.862234366242168e-07,
+      "loss": 0.1715,
+      "step": 2029
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6139827966690063,
+      "learning_rate": 3.767578426521923e-07,
+      "loss": 0.1655,
+      "step": 2030
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5823429226875305,
+      "learning_rate": 3.674092453598954e-07,
+      "loss": 0.1739,
+      "step": 2031
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.455282986164093,
+      "learning_rate": 3.5817766678850615e-07,
+      "loss": 0.1286,
+      "step": 2032
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.6837687492370605,
+      "learning_rate": 3.490631287033197e-07,
+      "loss": 0.2482,
+      "step": 2033
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.555550217628479,
+      "learning_rate": 3.4006565259366295e-07,
+      "loss": 0.1769,
+      "step": 2034
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5346921682357788,
+      "learning_rate": 3.3118525967289483e-07,
+      "loss": 0.1493,
+      "step": 2035
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.6123247146606445,
+      "learning_rate": 3.224219708782894e-07,
+      "loss": 0.223,
+      "step": 2036
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.4993426203727722,
+      "learning_rate": 3.1377580687106944e-07,
+      "loss": 0.1488,
+      "step": 2037
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.38787686824798584,
+      "learning_rate": 3.052467880362675e-07,
+      "loss": 0.1166,
+      "step": 2038
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.6245026588439941,
+      "learning_rate": 2.968349344827592e-07,
+      "loss": 0.1659,
+      "step": 2039
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5814539194107056,
+      "learning_rate": 2.88540266043158e-07,
+      "loss": 0.1826,
+      "step": 2040
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5621504783630371,
+      "learning_rate": 2.8036280227379806e-07,
+      "loss": 0.1843,
+      "step": 2041
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.730143129825592,
+      "learning_rate": 2.723025624546849e-07,
+      "loss": 0.2372,
+      "step": 2042
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.5334561467170715,
+      "learning_rate": 2.6435956558943374e-07,
+      "loss": 0.1521,
+      "step": 2043
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.6417514681816101,
+      "learning_rate": 2.5653383040524227e-07,
+      "loss": 0.2306,
+      "step": 2044
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.6519188284873962,
+      "learning_rate": 2.488253753528458e-07,
+      "loss": 0.2459,
+      "step": 2045
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.621303141117096,
+      "learning_rate": 2.412342186064565e-07,
+      "loss": 0.1926,
+      "step": 2046
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.586046576499939,
+      "learning_rate": 2.3376037806374097e-07,
+      "loss": 0.2311,
+      "step": 2047
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.6882225871086121,
+      "learning_rate": 2.2640387134577058e-07,
+      "loss": 0.2099,
+      "step": 2048
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.49546754360198975,
+      "learning_rate": 2.1916471579697117e-07,
+      "loss": 0.1549,
+      "step": 2049
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.6687613129615784,
+      "learning_rate": 2.1204292848509555e-07,
+      "loss": 0.1868,
+      "step": 2050
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.5582687258720398,
+      "learning_rate": 2.0503852620117892e-07,
+      "loss": 0.1765,
+      "step": 2051
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.4284742474555969,
+      "learning_rate": 1.98151525459489e-07,
+      "loss": 0.1803,
+      "step": 2052
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5616750121116638,
+      "learning_rate": 1.9138194249750386e-07,
+      "loss": 0.13,
+      "step": 2053
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4798294007778168,
+      "learning_rate": 1.8472979327587292e-07,
+      "loss": 0.1767,
+      "step": 2054
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5998563170433044,
+      "learning_rate": 1.781950934783505e-07,
+      "loss": 0.174,
+      "step": 2055
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.6948825120925903,
+      "learning_rate": 1.717778585118013e-07,
+      "loss": 0.2749,
+      "step": 2056
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.6227206587791443,
+      "learning_rate": 1.6547810350613368e-07,
+      "loss": 0.2044,
+      "step": 2057
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5002305507659912,
+      "learning_rate": 1.5929584331427216e-07,
+      "loss": 0.1208,
+      "step": 2058
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.44236671924591064,
+      "learning_rate": 1.5323109251212943e-07,
+      "loss": 0.1347,
+      "step": 2059
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4875430464744568,
+      "learning_rate": 1.4728386539856754e-07,
+      "loss": 0.1416,
+      "step": 2060
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5875696539878845,
+      "learning_rate": 1.4145417599534804e-07,
+      "loss": 0.1956,
+      "step": 2061
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.496926873922348,
+      "learning_rate": 1.3574203804713747e-07,
+      "loss": 0.1535,
+      "step": 2062
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5441425442695618,
+      "learning_rate": 1.3014746502142962e-07,
+      "loss": 0.1671,
+      "step": 2063
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.562794029712677,
+      "learning_rate": 1.246704701085566e-07,
+      "loss": 0.1922,
+      "step": 2064
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5756726264953613,
+      "learning_rate": 1.1931106622161124e-07,
+      "loss": 0.1744,
+      "step": 2065
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.602090060710907,
+      "learning_rate": 1.1406926599646373e-07,
+      "loss": 0.182,
+      "step": 2066
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.4644905924797058,
+      "learning_rate": 1.0894508179170038e-07,
+      "loss": 0.1557,
+      "step": 2067
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.35805532336235046,
+      "learning_rate": 1.0393852568860719e-07,
+      "loss": 0.1086,
+      "step": 2068
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.6840491890907288,
+      "learning_rate": 9.904960949114194e-08,
+      "loss": 0.2197,
+      "step": 2069
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.6050341129302979,
+      "learning_rate": 9.427834472588992e-08,
+      "loss": 0.2051,
+      "step": 2070
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.6657841205596924,
+      "learning_rate": 8.962474264206377e-08,
+      "loss": 0.2464,
+      "step": 2071
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5474058985710144,
+      "learning_rate": 8.508881421145365e-08,
+      "loss": 0.1659,
+      "step": 2072
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5595186352729797,
+      "learning_rate": 8.067057012842161e-08,
+      "loss": 0.1514,
+      "step": 2073
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6423659324645996,
+      "learning_rate": 7.637002080985168e-08,
+      "loss": 0.1948,
+      "step": 2074
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5785170197486877,
+      "learning_rate": 7.218717639514983e-08,
+      "loss": 0.209,
+      "step": 2075
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5037190914154053,
+      "learning_rate": 6.812204674619959e-08,
+      "loss": 0.1653,
+      "step": 2076
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5250243544578552,
+      "learning_rate": 6.417464144736208e-08,
+      "loss": 0.1821,
+      "step": 2077
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.4643983840942383,
+      "learning_rate": 6.034496980542037e-08,
+      "loss": 0.1409,
+      "step": 2078
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.45054012537002563,
+      "learning_rate": 5.663304084960186e-08,
+      "loss": 0.1308,
+      "step": 2079
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.7427161335945129,
+      "learning_rate": 5.303886333151153e-08,
+      "loss": 0.1643,
+      "step": 2080
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5470144748687744,
+      "learning_rate": 4.9562445725132026e-08,
+      "loss": 0.1944,
+      "step": 2081
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6634718179702759,
+      "learning_rate": 4.620379622682358e-08,
+      "loss": 0.2249,
+      "step": 2082
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5589112639427185,
+      "learning_rate": 4.296292275526859e-08,
+      "loss": 0.1445,
+      "step": 2083
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5131627321243286,
+      "learning_rate": 3.9839832951465986e-08,
+      "loss": 0.1988,
+      "step": 2084
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.6173979043960571,
+      "learning_rate": 3.6834534178725734e-08,
+      "loss": 0.2527,
+      "step": 2085
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5690194964408875,
+      "learning_rate": 3.394703352263551e-08,
+      "loss": 0.1965,
+      "step": 2086
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.457841694355011,
+      "learning_rate": 3.117733779105514e-08,
+      "loss": 0.1219,
+      "step": 2087
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5733069181442261,
+      "learning_rate": 2.8525453514099965e-08,
+      "loss": 0.1835,
+      "step": 2088
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5969994068145752,
+      "learning_rate": 2.599138694410752e-08,
+      "loss": 0.1975,
+      "step": 2089
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5808123350143433,
+      "learning_rate": 2.3575144055643094e-08,
+      "loss": 0.1766,
+      "step": 2090
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5589935183525085,
+      "learning_rate": 2.127673054548862e-08,
+      "loss": 0.1502,
+      "step": 2091
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.6381707191467285,
+      "learning_rate": 1.9096151832609375e-08,
+      "loss": 0.1961,
+      "step": 2092
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5051969885826111,
+      "learning_rate": 1.703341305815398e-08,
+      "loss": 0.1417,
+      "step": 2093
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.71206134557724,
+      "learning_rate": 1.5088519085437734e-08,
+      "loss": 0.2117,
+      "step": 2094
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5861583352088928,
+      "learning_rate": 1.3261474499937087e-08,
+      "loss": 0.2305,
+      "step": 2095
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5876367092132568,
+      "learning_rate": 1.1552283609272962e-08,
+      "loss": 0.193,
+      "step": 2096
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5396059155464172,
+      "learning_rate": 9.96095044320522e-09,
+      "loss": 0.184,
+      "step": 2097
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.6349846124649048,
+      "learning_rate": 8.487478753615997e-09,
+      "loss": 0.2258,
+      "step": 2098
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5600652694702148,
+      "learning_rate": 7.13187201450971e-09,
+      "loss": 0.1696,
+      "step": 2099
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.4990902543067932,
+      "learning_rate": 5.894133422001957e-09,
+      "loss": 0.155,
+      "step": 2100
+    },
+    {
+      "epoch": 1.99,
+      "eval_loss": 0.28876304626464844,
+      "eval_runtime": 364.7764,
+      "eval_samples_per_second": 0.935,
+      "eval_steps_per_second": 0.236,
+      "step": 2100
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.38263845443725586,
+      "learning_rate": 4.774265894302854e-09,
+      "loss": 0.104,
+      "step": 2101
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.4471862316131592,
+      "learning_rate": 3.772272071722593e-09,
+      "loss": 0.1028,
+      "step": 2102
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5412765741348267,
+      "learning_rate": 2.888154316671443e-09,
+      "loss": 0.1408,
+      "step": 2103
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5564787983894348,
+      "learning_rate": 2.1219147136264382e-09,
+      "loss": 0.1578,
+      "step": 2104
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5173894762992859,
+      "learning_rate": 1.473555069148036e-09,
+      "loss": 0.1767,
+      "step": 2105
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5809282064437866,
+      "learning_rate": 9.43076911874563e-10,
+      "loss": 0.1415,
+      "step": 2106
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.46766799688339233,
+      "learning_rate": 5.304814925111146e-10,
+      "loss": 0.1206,
+      "step": 2107
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6120114922523499,
+      "learning_rate": 2.3576978384065584e-10,
+      "loss": 0.2087,
+      "step": 2108
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5634140372276306,
+      "learning_rate": 5.894248070181796e-11,
+      "loss": 0.1613,
+      "step": 2109
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.4470638632774353,
+      "learning_rate": 0.0,
+      "loss": 0.1331,
+      "step": 2110
+    },
+    {
+      "epoch": 2.0,
+      "step": 2110,
+      "total_flos": 2.318442573993083e+17,
+      "train_loss": 0.25298224955232224,
+      "train_runtime": 111762.6633,
+      "train_samples_per_second": 0.604,
+      "train_steps_per_second": 0.019
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2110,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 100,
+  "total_flos": 2.318442573993083e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}