diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/README.md b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..d4576fe074287232d3836bf69c21d3f2593290d9
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_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_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/adapter_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9417461a6e039c4ab9fad59760d9fce08b4f020c
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_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-qwen2-0.5b-ov/",
+  "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": [
+    "gate_proj",
+    "down_proj",
+    "k_proj",
+    "up_proj",
+    "v_proj",
+    "o_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}
\ No newline at end of file
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/adapter_model.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..fc26e0c41184601705bb4f680b7da9fc2489fce8
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0f9fb98dc1db5d22e2d679095473206f18ff3a19f0549872317fb6d5943c7f9e
+size 186956026
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..478f1491b67376978541c36f238c34d5e5db496d
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/config.json
@@ -0,0 +1,68 @@
+{
+  "_name_or_path": "/data/wiedmann/hub/models--lmms-lab--llava-onevision-qwen2-0.5b-ov/",
+  "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": 896,
+  "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": 4864,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 24,
+  "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": 2,
+  "mm_streaming_num_state_tokens": 2048,
+  "mm_streaming_patches_per_frame": 729,
+  "mm_streaming_state_dim": 1152,
+  "mm_streaming_vision_chunk_size": 4,
+  "mm_tunable_parts": "mm_mlp_adapter,mm_vision_resampler",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "mm_vision_select_feature": "patch",
+  "mm_vision_select_layer": -2,
+  "mm_vision_tower": "google/siglip-so400m-patch14-384",
+  "mm_vision_tower_lr": null,
+  "model_type": "llava_qwen",
+  "num_attention_heads": 14,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 2,
+  "pos_skipping_range": 4096,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "tie_word_embeddings": true,
+  "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": 151936
+}
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/generation_config.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..0b6ab6dabdd2b7e2b8434b1a151deeff32eb04e5
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_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": 1.1,
+  "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_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/non_lora_trainables.bin b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..368811e3cb809050b722be1702029218b3c25810
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7003d5230cd01afee57d0689233d54caecd51b0cedf6801d1646585122e758b0
+size 181159642
diff --git a/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/trainer_state.json b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..56ab87dc8b5f61c41de44173d91b1a13d0fa4b91
--- /dev/null
+++ b/llavanext-google_siglip-so400m-patch14-384-Qwen_Qwen2-0.5B-Instruct-nextqa_shuffled-streaming_baseline_dualdecode_lora_agg_fc4_s2048/trainer_state.json
@@ -0,0 +1,7495 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9996446760630108,
+  "eval_steps": 100,
+  "global_step": 1055,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.097281455993652,
+      "learning_rate": 3.125e-06,
+      "loss": 0.7487,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 16.0189266204834,
+      "learning_rate": 6.25e-06,
+      "loss": 0.8455,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 16.745656967163086,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 1.0149,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.859532356262207,
+      "learning_rate": 1.25e-05,
+      "loss": 0.8081,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.44382905960083,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.5189,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.116035461425781,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.5057,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.0932793617248535,
+      "learning_rate": 2.1875e-05,
+      "loss": 0.5296,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.945772647857666,
+      "learning_rate": 2.5e-05,
+      "loss": 0.3803,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7893362045288086,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 0.3804,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.049710988998413,
+      "learning_rate": 3.125e-05,
+      "loss": 0.3705,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.4781906604766846,
+      "learning_rate": 3.4375e-05,
+      "loss": 0.3579,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.9901652336120605,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.326,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.247368574142456,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.3797,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.500739097595215,
+      "learning_rate": 4.375e-05,
+      "loss": 0.4391,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.295069694519043,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.5952,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.069502115249634,
+      "learning_rate": 5e-05,
+      "loss": 0.4515,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.1711320877075195,
+      "learning_rate": 5.3125000000000004e-05,
+      "loss": 0.5197,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8819308280944824,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 0.3522,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8062694072723389,
+      "learning_rate": 5.9375e-05,
+      "loss": 0.421,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1293818950653076,
+      "learning_rate": 6.25e-05,
+      "loss": 0.2876,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6478607654571533,
+      "learning_rate": 6.562500000000001e-05,
+      "loss": 0.4442,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3513648509979248,
+      "learning_rate": 6.875e-05,
+      "loss": 0.213,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5565677881240845,
+      "learning_rate": 7.1875e-05,
+      "loss": 0.3601,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.745936393737793,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.5552,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6410375833511353,
+      "learning_rate": 7.8125e-05,
+      "loss": 0.2627,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7321958541870117,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 0.4067,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8782188892364502,
+      "learning_rate": 8.4375e-05,
+      "loss": 0.447,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8769721984863281,
+      "learning_rate": 8.75e-05,
+      "loss": 0.3918,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.514636754989624,
+      "learning_rate": 9.062500000000001e-05,
+      "loss": 0.3864,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4011240005493164,
+      "learning_rate": 9.375e-05,
+      "loss": 0.3248,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.751269817352295,
+      "learning_rate": 9.687500000000001e-05,
+      "loss": 0.4012,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.7858824729919434,
+      "learning_rate": 0.0001,
+      "loss": 0.3511,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8888771533966064,
+      "learning_rate": 9.999976423021617e-05,
+      "loss": 0.4655,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.5688952207565308,
+      "learning_rate": 9.999905692308813e-05,
+      "loss": 0.4285,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8022360801696777,
+      "learning_rate": 9.999787808528638e-05,
+      "loss": 0.5391,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.214371681213379,
+      "learning_rate": 9.999622772792828e-05,
+      "loss": 0.3455,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.461837649345398,
+      "learning_rate": 9.9994105866578e-05,
+      "loss": 0.4192,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.6857012510299683,
+      "learning_rate": 9.999151252124639e-05,
+      "loss": 0.4272,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1822096109390259,
+      "learning_rate": 9.998844771639073e-05,
+      "loss": 0.3249,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2399797439575195,
+      "learning_rate": 9.998491148091457e-05,
+      "loss": 0.3635,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4649162292480469,
+      "learning_rate": 9.99809038481674e-05,
+      "loss": 0.4571,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5287866592407227,
+      "learning_rate": 9.997642485594436e-05,
+      "loss": 0.4268,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3567689657211304,
+      "learning_rate": 9.99714745464859e-05,
+      "loss": 0.384,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.2080179452896118,
+      "learning_rate": 9.996605296647737e-05,
+      "loss": 0.3516,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.531796932220459,
+      "learning_rate": 9.996016016704854e-05,
+      "loss": 0.5773,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1242033243179321,
+      "learning_rate": 9.995379620377318e-05,
+      "loss": 0.3346,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0072808265686035,
+      "learning_rate": 9.99469611366685e-05,
+      "loss": 0.3377,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.274427056312561,
+      "learning_rate": 9.993965503019457e-05,
+      "loss": 0.4925,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2531381845474243,
+      "learning_rate": 9.99318779532538e-05,
+      "loss": 0.3972,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1453968286514282,
+      "learning_rate": 9.992362997919016e-05,
+      "loss": 0.3751,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1725373268127441,
+      "learning_rate": 9.991491118578856e-05,
+      "loss": 0.4297,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1341646909713745,
+      "learning_rate": 9.990572165527413e-05,
+      "loss": 0.409,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3306612968444824,
+      "learning_rate": 9.98960614743114e-05,
+      "loss": 0.4996,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3339848518371582,
+      "learning_rate": 9.988593073400354e-05,
+      "loss": 0.5204,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3567358255386353,
+      "learning_rate": 9.987532952989145e-05,
+      "loss": 0.4268,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0651249885559082,
+      "learning_rate": 9.986425796195287e-05,
+      "loss": 0.3347,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2171188592910767,
+      "learning_rate": 9.985271613460144e-05,
+      "loss": 0.4888,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.9910860061645508,
+      "learning_rate": 9.984070415668574e-05,
+      "loss": 0.3146,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2523832321166992,
+      "learning_rate": 9.982822214148821e-05,
+      "loss": 0.4405,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9510688185691833,
+      "learning_rate": 9.981527020672413e-05,
+      "loss": 0.3701,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2030282020568848,
+      "learning_rate": 9.980184847454052e-05,
+      "loss": 0.4541,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0943279266357422,
+      "learning_rate": 9.978795707151492e-05,
+      "loss": 0.3863,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.121793270111084,
+      "learning_rate": 9.977359612865423e-05,
+      "loss": 0.3421,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1635420322418213,
+      "learning_rate": 9.975876578139355e-05,
+      "loss": 0.3801,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2730247974395752,
+      "learning_rate": 9.974346616959476e-05,
+      "loss": 0.5326,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.185283899307251,
+      "learning_rate": 9.972769743754531e-05,
+      "loss": 0.3863,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9916014075279236,
+      "learning_rate": 9.971145973395684e-05,
+      "loss": 0.2977,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.134338617324829,
+      "learning_rate": 9.969475321196373e-05,
+      "loss": 0.2865,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2008209228515625,
+      "learning_rate": 9.96775780291217e-05,
+      "loss": 0.3408,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1887668371200562,
+      "learning_rate": 9.965993434740633e-05,
+      "loss": 0.3459,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.357417106628418,
+      "learning_rate": 9.96418223332115e-05,
+      "loss": 0.4156,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0568681955337524,
+      "learning_rate": 9.962324215734782e-05,
+      "loss": 0.2776,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3621985912322998,
+      "learning_rate": 9.960419399504106e-05,
+      "loss": 0.448,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0651113986968994,
+      "learning_rate": 9.958467802593046e-05,
+      "loss": 0.3291,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2189366817474365,
+      "learning_rate": 9.956469443406706e-05,
+      "loss": 0.3933,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0380326509475708,
+      "learning_rate": 9.954424340791196e-05,
+      "loss": 0.334,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.1057391166687012,
+      "learning_rate": 9.952332514033447e-05,
+      "loss": 0.3834,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.187192440032959,
+      "learning_rate": 9.950193982861047e-05,
+      "loss": 0.454,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.14493989944458,
+      "learning_rate": 9.948008767442034e-05,
+      "loss": 0.393,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9716129302978516,
+      "learning_rate": 9.94577688838472e-05,
+      "loss": 0.3315,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1064670085906982,
+      "learning_rate": 9.943498366737487e-05,
+      "loss": 0.407,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0029715299606323,
+      "learning_rate": 9.941173223988602e-05,
+      "loss": 0.3627,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1206356287002563,
+      "learning_rate": 9.938801482065998e-05,
+      "loss": 0.4091,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.167804479598999,
+      "learning_rate": 9.93638316333708e-05,
+      "loss": 0.4626,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.018316626548767,
+      "learning_rate": 9.933918290608509e-05,
+      "loss": 0.3882,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0729396343231201,
+      "learning_rate": 9.93140688712598e-05,
+      "loss": 0.4301,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.225366234779358,
+      "learning_rate": 9.928848976574019e-05,
+      "loss": 0.5055,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1701208353042603,
+      "learning_rate": 9.926244583075748e-05,
+      "loss": 0.4201,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.1674448251724243,
+      "learning_rate": 9.923593731192655e-05,
+      "loss": 0.4324,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0377925634384155,
+      "learning_rate": 9.920896445924373e-05,
+      "loss": 0.3535,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9072185158729553,
+      "learning_rate": 9.918152752708438e-05,
+      "loss": 0.256,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.2062162160873413,
+      "learning_rate": 9.915362677420044e-05,
+      "loss": 0.4453,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.3401453495025635,
+      "learning_rate": 9.912526246371814e-05,
+      "loss": 0.4987,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0644365549087524,
+      "learning_rate": 9.909643486313533e-05,
+      "loss": 0.4206,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1782938241958618,
+      "learning_rate": 9.906714424431913e-05,
+      "loss": 0.4671,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5179443359375,
+      "learning_rate": 9.903739088350325e-05,
+      "loss": 0.3056,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0053291320800781,
+      "learning_rate": 9.900717506128541e-05,
+      "loss": 0.3314,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.0548299551010132,
+      "learning_rate": 9.897649706262473e-05,
+      "loss": 0.3612,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1142017841339111,
+      "learning_rate": 9.894535717683902e-05,
+      "loss": 0.4241,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9980804920196533,
+      "learning_rate": 9.891375569760204e-05,
+      "loss": 0.4129,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.36958399415016174,
+      "eval_runtime": 364.8768,
+      "eval_samples_per_second": 0.935,
+      "eval_steps_per_second": 0.236,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.986346423625946,
+      "learning_rate": 9.888169292294076e-05,
+      "loss": 0.3537,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0398842096328735,
+      "learning_rate": 9.884916915523251e-05,
+      "loss": 0.3339,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.124420166015625,
+      "learning_rate": 9.881618470120215e-05,
+      "loss": 0.4144,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0602110624313354,
+      "learning_rate": 9.87827398719192e-05,
+      "loss": 0.3897,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0206547975540161,
+      "learning_rate": 9.874883498279484e-05,
+      "loss": 0.3295,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9855296015739441,
+      "learning_rate": 9.871447035357903e-05,
+      "loss": 0.3983,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0601434707641602,
+      "learning_rate": 9.867964630835743e-05,
+      "loss": 0.3573,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0047048330307007,
+      "learning_rate": 9.864436317554831e-05,
+      "loss": 0.3455,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9392563700675964,
+      "learning_rate": 9.860862128789953e-05,
+      "loss": 0.3401,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.0609445571899414,
+      "learning_rate": 9.857242098248542e-05,
+      "loss": 0.2994,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9969935417175293,
+      "learning_rate": 9.853576260070347e-05,
+      "loss": 0.3885,
+      "step": 111
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8490005731582642,
+      "learning_rate": 9.849864648827125e-05,
+      "loss": 0.3419,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0373643636703491,
+      "learning_rate": 9.846107299522304e-05,
+      "loss": 0.4178,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.1603271961212158,
+      "learning_rate": 9.842304247590667e-05,
+      "loss": 0.5297,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8847191333770752,
+      "learning_rate": 9.838455528897998e-05,
+      "loss": 0.3255,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9143794178962708,
+      "learning_rate": 9.834561179740762e-05,
+      "loss": 0.3473,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0277754068374634,
+      "learning_rate": 9.830621236845754e-05,
+      "loss": 0.3736,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8876435160636902,
+      "learning_rate": 9.826635737369751e-05,
+      "loss": 0.367,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0674887895584106,
+      "learning_rate": 9.82260471889917e-05,
+      "loss": 0.3307,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8935846090316772,
+      "learning_rate": 9.818528219449705e-05,
+      "loss": 0.3287,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8292754888534546,
+      "learning_rate": 9.814406277465968e-05,
+      "loss": 0.2848,
+      "step": 121
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.982430636882782,
+      "learning_rate": 9.810238931821138e-05,
+      "loss": 0.3762,
+      "step": 122
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9691665172576904,
+      "learning_rate": 9.806026221816581e-05,
+      "loss": 0.4329,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.047287106513977,
+      "learning_rate": 9.801768187181486e-05,
+      "loss": 0.4535,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9385687112808228,
+      "learning_rate": 9.797464868072488e-05,
+      "loss": 0.3351,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9173955321311951,
+      "learning_rate": 9.793116305073292e-05,
+      "loss": 0.3607,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9677826166152954,
+      "learning_rate": 9.788722539194291e-05,
+      "loss": 0.4076,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9736503958702087,
+      "learning_rate": 9.784283611872169e-05,
+      "loss": 0.3777,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1609407663345337,
+      "learning_rate": 9.77979956496953e-05,
+      "loss": 0.3964,
+      "step": 129
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.083729863166809,
+      "learning_rate": 9.77527044077448e-05,
+      "loss": 0.5007,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.038608193397522,
+      "learning_rate": 9.770696282000244e-05,
+      "loss": 0.4177,
+      "step": 131
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0175806283950806,
+      "learning_rate": 9.766077131784763e-05,
+      "loss": 0.4756,
+      "step": 132
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9377140402793884,
+      "learning_rate": 9.761413033690276e-05,
+      "loss": 0.3628,
+      "step": 133
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9915252923965454,
+      "learning_rate": 9.756704031702918e-05,
+      "loss": 0.4229,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9683788418769836,
+      "learning_rate": 9.751950170232309e-05,
+      "loss": 0.3578,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9741252660751343,
+      "learning_rate": 9.74715149411112e-05,
+      "loss": 0.3975,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.9931705594062805,
+      "learning_rate": 9.742308048594665e-05,
+      "loss": 0.4423,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7277801632881165,
+      "learning_rate": 9.73741987936047e-05,
+      "loss": 0.2263,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7658216953277588,
+      "learning_rate": 9.732487032507837e-05,
+      "loss": 0.2672,
+      "step": 139
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1071852445602417,
+      "learning_rate": 9.727509554557417e-05,
+      "loss": 0.4585,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8707927465438843,
+      "learning_rate": 9.722487492450764e-05,
+      "loss": 0.3335,
+      "step": 141
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.0920414924621582,
+      "learning_rate": 9.717420893549902e-05,
+      "loss": 0.4208,
+      "step": 142
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0512176752090454,
+      "learning_rate": 9.712309805636863e-05,
+      "loss": 0.3804,
+      "step": 143
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.1570490598678589,
+      "learning_rate": 9.707154276913256e-05,
+      "loss": 0.4617,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0212880373001099,
+      "learning_rate": 9.701954355999792e-05,
+      "loss": 0.4306,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9789423942565918,
+      "learning_rate": 9.696710091935841e-05,
+      "loss": 0.3941,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9384461641311646,
+      "learning_rate": 9.691421534178966e-05,
+      "loss": 0.3574,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9509838819503784,
+      "learning_rate": 9.686088732604449e-05,
+      "loss": 0.3708,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8480948805809021,
+      "learning_rate": 9.68071173750483e-05,
+      "loss": 0.2689,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0354323387145996,
+      "learning_rate": 9.675290599589429e-05,
+      "loss": 0.4639,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.2338851690292358,
+      "learning_rate": 9.669825369983865e-05,
+      "loss": 0.3605,
+      "step": 151
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8615697622299194,
+      "learning_rate": 9.664316100229578e-05,
+      "loss": 0.3979,
+      "step": 152
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.0251574516296387,
+      "learning_rate": 9.658762842283343e-05,
+      "loss": 0.4112,
+      "step": 153
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8077225089073181,
+      "learning_rate": 9.653165648516776e-05,
+      "loss": 0.3142,
+      "step": 154
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9959987998008728,
+      "learning_rate": 9.647524571715843e-05,
+      "loss": 0.408,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.0971671342849731,
+      "learning_rate": 9.641839665080363e-05,
+      "loss": 0.4065,
+      "step": 156
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9865227937698364,
+      "learning_rate": 9.636110982223505e-05,
+      "loss": 0.3712,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.007011890411377,
+      "learning_rate": 9.630338577171282e-05,
+      "loss": 0.3979,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9343575239181519,
+      "learning_rate": 9.624522504362039e-05,
+      "loss": 0.3449,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8832029700279236,
+      "learning_rate": 9.618662818645948e-05,
+      "loss": 0.4001,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7331508994102478,
+      "learning_rate": 9.612759575284483e-05,
+      "loss": 0.2553,
+      "step": 161
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9810391664505005,
+      "learning_rate": 9.606812829949896e-05,
+      "loss": 0.3842,
+      "step": 162
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8460718989372253,
+      "learning_rate": 9.600822638724705e-05,
+      "loss": 0.3192,
+      "step": 163
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8436979055404663,
+      "learning_rate": 9.594789058101153e-05,
+      "loss": 0.3114,
+      "step": 164
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8543170690536499,
+      "learning_rate": 9.588712144980681e-05,
+      "loss": 0.3758,
+      "step": 165
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0378081798553467,
+      "learning_rate": 9.582591956673387e-05,
+      "loss": 0.3823,
+      "step": 166
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9082635641098022,
+      "learning_rate": 9.576428550897489e-05,
+      "loss": 0.3048,
+      "step": 167
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1705089807510376,
+      "learning_rate": 9.570221985778785e-05,
+      "loss": 0.4402,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9553974866867065,
+      "learning_rate": 9.563972319850092e-05,
+      "loss": 0.3585,
+      "step": 169
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7092993259429932,
+      "learning_rate": 9.557679612050708e-05,
+      "loss": 0.2339,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8900098204612732,
+      "learning_rate": 9.551343921725843e-05,
+      "loss": 0.4041,
+      "step": 171
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.0518604516983032,
+      "learning_rate": 9.544965308626074e-05,
+      "loss": 0.497,
+      "step": 172
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8250086903572083,
+      "learning_rate": 9.538543832906773e-05,
+      "loss": 0.2985,
+      "step": 173
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9862067699432373,
+      "learning_rate": 9.53207955512753e-05,
+      "loss": 0.3648,
+      "step": 174
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.162681221961975,
+      "learning_rate": 9.525572536251607e-05,
+      "loss": 0.4936,
+      "step": 175
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9331067800521851,
+      "learning_rate": 9.519022837645337e-05,
+      "loss": 0.3807,
+      "step": 176
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9334955811500549,
+      "learning_rate": 9.512430521077565e-05,
+      "loss": 0.3719,
+      "step": 177
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0926851034164429,
+      "learning_rate": 9.505795648719048e-05,
+      "loss": 0.4157,
+      "step": 178
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0729618072509766,
+      "learning_rate": 9.499118283141887e-05,
+      "loss": 0.456,
+      "step": 179
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9267973899841309,
+      "learning_rate": 9.492398487318921e-05,
+      "loss": 0.3834,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8041082620620728,
+      "learning_rate": 9.485636324623146e-05,
+      "loss": 0.2984,
+      "step": 181
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7916663885116577,
+      "learning_rate": 9.478831858827104e-05,
+      "loss": 0.3551,
+      "step": 182
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8479263186454773,
+      "learning_rate": 9.471985154102292e-05,
+      "loss": 0.3749,
+      "step": 183
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.015342116355896,
+      "learning_rate": 9.465096275018556e-05,
+      "loss": 0.4887,
+      "step": 184
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8462713360786438,
+      "learning_rate": 9.458165286543476e-05,
+      "loss": 0.3611,
+      "step": 185
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0650149583816528,
+      "learning_rate": 9.451192254041758e-05,
+      "loss": 0.424,
+      "step": 186
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0671846866607666,
+      "learning_rate": 9.444177243274618e-05,
+      "loss": 0.5405,
+      "step": 187
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9743732810020447,
+      "learning_rate": 9.437120320399158e-05,
+      "loss": 0.4259,
+      "step": 188
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8786170482635498,
+      "learning_rate": 9.430021551967744e-05,
+      "loss": 0.2968,
+      "step": 189
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9776631593704224,
+      "learning_rate": 9.422881004927381e-05,
+      "loss": 0.3411,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.9059628248214722,
+      "learning_rate": 9.415698746619079e-05,
+      "loss": 0.3931,
+      "step": 191
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7348672747612,
+      "learning_rate": 9.408474844777217e-05,
+      "loss": 0.2788,
+      "step": 192
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0708872079849243,
+      "learning_rate": 9.401209367528908e-05,
+      "loss": 0.5358,
+      "step": 193
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8517162203788757,
+      "learning_rate": 9.393902383393347e-05,
+      "loss": 0.3604,
+      "step": 194
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.144752860069275,
+      "learning_rate": 9.386553961281179e-05,
+      "loss": 0.5402,
+      "step": 195
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8977828621864319,
+      "learning_rate": 9.379164170493844e-05,
+      "loss": 0.4353,
+      "step": 196
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9361101984977722,
+      "learning_rate": 9.371733080722911e-05,
+      "loss": 0.3933,
+      "step": 197
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7996360063552856,
+      "learning_rate": 9.36426076204944e-05,
+      "loss": 0.2878,
+      "step": 198
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8819470405578613,
+      "learning_rate": 9.35674728494331e-05,
+      "loss": 0.4004,
+      "step": 199
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8699619770050049,
+      "learning_rate": 9.349192720262555e-05,
+      "loss": 0.3392,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.3580893576145172,
+      "eval_runtime": 364.0536,
+      "eval_samples_per_second": 0.937,
+      "eval_steps_per_second": 0.236,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9195945858955383,
+      "learning_rate": 9.341597139252698e-05,
+      "loss": 0.395,
+      "step": 201
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9107511043548584,
+      "learning_rate": 9.333960613546079e-05,
+      "loss": 0.3825,
+      "step": 202
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7383326888084412,
+      "learning_rate": 9.326283215161178e-05,
+      "loss": 0.2702,
+      "step": 203
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7380225658416748,
+      "learning_rate": 9.31856501650194e-05,
+      "loss": 0.1995,
+      "step": 204
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8848481774330139,
+      "learning_rate": 9.310806090357082e-05,
+      "loss": 0.4128,
+      "step": 205
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7994688153266907,
+      "learning_rate": 9.303006509899418e-05,
+      "loss": 0.228,
+      "step": 206
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9198280572891235,
+      "learning_rate": 9.29516634868517e-05,
+      "loss": 0.3824,
+      "step": 207
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9306249618530273,
+      "learning_rate": 9.287285680653254e-05,
+      "loss": 0.4286,
+      "step": 208
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8849806189537048,
+      "learning_rate": 9.279364580124614e-05,
+      "loss": 0.3411,
+      "step": 209
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7815821766853333,
+      "learning_rate": 9.27140312180149e-05,
+      "loss": 0.2701,
+      "step": 210
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0098775625228882,
+      "learning_rate": 9.263401380766739e-05,
+      "loss": 0.4615,
+      "step": 211
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.872482419013977,
+      "learning_rate": 9.255359432483105e-05,
+      "loss": 0.3794,
+      "step": 212
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8410907983779907,
+      "learning_rate": 9.247277352792533e-05,
+      "loss": 0.3048,
+      "step": 213
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8730826377868652,
+      "learning_rate": 9.239155217915422e-05,
+      "loss": 0.4258,
+      "step": 214
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6687842607498169,
+      "learning_rate": 9.230993104449939e-05,
+      "loss": 0.2602,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7946191430091858,
+      "learning_rate": 9.222791089371265e-05,
+      "loss": 0.3194,
+      "step": 216
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7969677448272705,
+      "learning_rate": 9.214549250030899e-05,
+      "loss": 0.3208,
+      "step": 217
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8837888836860657,
+      "learning_rate": 9.206267664155907e-05,
+      "loss": 0.4715,
+      "step": 218
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7196596264839172,
+      "learning_rate": 9.197946409848194e-05,
+      "loss": 0.2804,
+      "step": 219
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.788112461566925,
+      "learning_rate": 9.189585565583778e-05,
+      "loss": 0.337,
+      "step": 220
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8203721642494202,
+      "learning_rate": 9.181185210212034e-05,
+      "loss": 0.387,
+      "step": 221
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7958011627197266,
+      "learning_rate": 9.172745422954961e-05,
+      "loss": 0.3246,
+      "step": 222
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0959714651107788,
+      "learning_rate": 9.164266283406433e-05,
+      "loss": 0.4353,
+      "step": 223
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.916005551815033,
+      "learning_rate": 9.155747871531444e-05,
+      "loss": 0.38,
+      "step": 224
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9572837948799133,
+      "learning_rate": 9.14719026766536e-05,
+      "loss": 0.4176,
+      "step": 225
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8459186553955078,
+      "learning_rate": 9.13859355251316e-05,
+      "loss": 0.3557,
+      "step": 226
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8369545936584473,
+      "learning_rate": 9.129957807148666e-05,
+      "loss": 0.3124,
+      "step": 227
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0042897462844849,
+      "learning_rate": 9.121283113013795e-05,
+      "loss": 0.4506,
+      "step": 228
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7852050065994263,
+      "learning_rate": 9.112569551917774e-05,
+      "loss": 0.2913,
+      "step": 229
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9431066513061523,
+      "learning_rate": 9.103817206036382e-05,
+      "loss": 0.4144,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.9962843656539917,
+      "learning_rate": 9.095026157911166e-05,
+      "loss": 0.4193,
+      "step": 231
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8542195558547974,
+      "learning_rate": 9.086196490448668e-05,
+      "loss": 0.3078,
+      "step": 232
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.2066110372543335,
+      "learning_rate": 9.077328286919638e-05,
+      "loss": 0.4059,
+      "step": 233
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.946294367313385,
+      "learning_rate": 9.068421630958254e-05,
+      "loss": 0.4201,
+      "step": 234
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7656259536743164,
+      "learning_rate": 9.059476606561328e-05,
+      "loss": 0.2675,
+      "step": 235
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8061059713363647,
+      "learning_rate": 9.050493298087523e-05,
+      "loss": 0.3768,
+      "step": 236
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8254323601722717,
+      "learning_rate": 9.041471790256542e-05,
+      "loss": 0.3771,
+      "step": 237
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8439212441444397,
+      "learning_rate": 9.032412168148345e-05,
+      "loss": 0.381,
+      "step": 238
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.0162991285324097,
+      "learning_rate": 9.02331451720234e-05,
+      "loss": 0.3881,
+      "step": 239
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.743945837020874,
+      "learning_rate": 9.014178923216572e-05,
+      "loss": 0.305,
+      "step": 240
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8028184771537781,
+      "learning_rate": 9.005005472346924e-05,
+      "loss": 0.2984,
+      "step": 241
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9433017373085022,
+      "learning_rate": 8.995794251106295e-05,
+      "loss": 0.4547,
+      "step": 242
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9029564261436462,
+      "learning_rate": 8.986545346363792e-05,
+      "loss": 0.36,
+      "step": 243
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7711244821548462,
+      "learning_rate": 8.977258845343905e-05,
+      "loss": 0.3814,
+      "step": 244
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9967159032821655,
+      "learning_rate": 8.967934835625689e-05,
+      "loss": 0.3588,
+      "step": 245
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9633325934410095,
+      "learning_rate": 8.958573405141932e-05,
+      "loss": 0.3841,
+      "step": 246
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8627241849899292,
+      "learning_rate": 8.949174642178332e-05,
+      "loss": 0.4477,
+      "step": 247
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9397616386413574,
+      "learning_rate": 8.939738635372664e-05,
+      "loss": 0.3797,
+      "step": 248
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.0576297044754028,
+      "learning_rate": 8.930265473713938e-05,
+      "loss": 0.4256,
+      "step": 249
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8548952341079712,
+      "learning_rate": 8.920755246541563e-05,
+      "loss": 0.4302,
+      "step": 250
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7654826641082764,
+      "learning_rate": 8.911208043544513e-05,
+      "loss": 0.3489,
+      "step": 251
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7768643498420715,
+      "learning_rate": 8.90162395476046e-05,
+      "loss": 0.3511,
+      "step": 252
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1376627683639526,
+      "learning_rate": 8.89200307057495e-05,
+      "loss": 0.5636,
+      "step": 253
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8920099139213562,
+      "learning_rate": 8.882345481720533e-05,
+      "loss": 0.4547,
+      "step": 254
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7642688155174255,
+      "learning_rate": 8.872651279275916e-05,
+      "loss": 0.3731,
+      "step": 255
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9503885507583618,
+      "learning_rate": 8.862920554665098e-05,
+      "loss": 0.3885,
+      "step": 256
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.8499652743339539,
+      "learning_rate": 8.853153399656513e-05,
+      "loss": 0.3722,
+      "step": 257
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9205160140991211,
+      "learning_rate": 8.843349906362163e-05,
+      "loss": 0.4551,
+      "step": 258
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8169696927070618,
+      "learning_rate": 8.833510167236747e-05,
+      "loss": 0.4371,
+      "step": 259
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7479214668273926,
+      "learning_rate": 8.823634275076791e-05,
+      "loss": 0.3615,
+      "step": 260
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8196630477905273,
+      "learning_rate": 8.813722323019773e-05,
+      "loss": 0.4328,
+      "step": 261
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6670660972595215,
+      "learning_rate": 8.803774404543247e-05,
+      "loss": 0.2887,
+      "step": 262
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7699260711669922,
+      "learning_rate": 8.793790613463955e-05,
+      "loss": 0.3748,
+      "step": 263
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5770822167396545,
+      "learning_rate": 8.783771043936948e-05,
+      "loss": 0.2356,
+      "step": 264
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8001684546470642,
+      "learning_rate": 8.773715790454695e-05,
+      "loss": 0.365,
+      "step": 265
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7842224836349487,
+      "learning_rate": 8.763624947846195e-05,
+      "loss": 0.2711,
+      "step": 266
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8602216243743896,
+      "learning_rate": 8.753498611276079e-05,
+      "loss": 0.3632,
+      "step": 267
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.955125629901886,
+      "learning_rate": 8.743336876243712e-05,
+      "loss": 0.441,
+      "step": 268
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.2633311748504639,
+      "learning_rate": 8.7331398385823e-05,
+      "loss": 0.5303,
+      "step": 269
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.0701488256454468,
+      "learning_rate": 8.722907594457974e-05,
+      "loss": 0.4324,
+      "step": 270
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.9061278104782104,
+      "learning_rate": 8.712640240368899e-05,
+      "loss": 0.4309,
+      "step": 271
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8782324194908142,
+      "learning_rate": 8.702337873144343e-05,
+      "loss": 0.3806,
+      "step": 272
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7344905734062195,
+      "learning_rate": 8.692000589943785e-05,
+      "loss": 0.2957,
+      "step": 273
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.016125202178955,
+      "learning_rate": 8.681628488255986e-05,
+      "loss": 0.4326,
+      "step": 274
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7400128245353699,
+      "learning_rate": 8.671221665898073e-05,
+      "loss": 0.3177,
+      "step": 275
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8918482661247253,
+      "learning_rate": 8.660780221014616e-05,
+      "loss": 0.3915,
+      "step": 276
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8788198232650757,
+      "learning_rate": 8.650304252076703e-05,
+      "loss": 0.3513,
+      "step": 277
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7801429629325867,
+      "learning_rate": 8.639793857881011e-05,
+      "loss": 0.3145,
+      "step": 278
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.9436314702033997,
+      "learning_rate": 8.629249137548874e-05,
+      "loss": 0.4672,
+      "step": 279
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8854785561561584,
+      "learning_rate": 8.618670190525352e-05,
+      "loss": 0.317,
+      "step": 280
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7875430583953857,
+      "learning_rate": 8.608057116578283e-05,
+      "loss": 0.3173,
+      "step": 281
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0251414775848389,
+      "learning_rate": 8.597410015797358e-05,
+      "loss": 0.4616,
+      "step": 282
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.85435551404953,
+      "learning_rate": 8.586728988593158e-05,
+      "loss": 0.2854,
+      "step": 283
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8404312133789062,
+      "learning_rate": 8.576014135696226e-05,
+      "loss": 0.35,
+      "step": 284
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8116161823272705,
+      "learning_rate": 8.565265558156101e-05,
+      "loss": 0.3202,
+      "step": 285
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.9451015591621399,
+      "learning_rate": 8.554483357340379e-05,
+      "loss": 0.3888,
+      "step": 286
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8051944971084595,
+      "learning_rate": 8.543667634933742e-05,
+      "loss": 0.2969,
+      "step": 287
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8410780429840088,
+      "learning_rate": 8.532818492937014e-05,
+      "loss": 0.3469,
+      "step": 288
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8163617253303528,
+      "learning_rate": 8.521936033666187e-05,
+      "loss": 0.3691,
+      "step": 289
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8933125734329224,
+      "learning_rate": 8.511020359751466e-05,
+      "loss": 0.3337,
+      "step": 290
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.689122200012207,
+      "learning_rate": 8.500071574136295e-05,
+      "loss": 0.2468,
+      "step": 291
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.845005989074707,
+      "learning_rate": 8.489089780076386e-05,
+      "loss": 0.3437,
+      "step": 292
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7403581142425537,
+      "learning_rate": 8.478075081138745e-05,
+      "loss": 0.3306,
+      "step": 293
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.955531656742096,
+      "learning_rate": 8.467027581200702e-05,
+      "loss": 0.4955,
+      "step": 294
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.0030899047851562,
+      "learning_rate": 8.455947384448924e-05,
+      "loss": 0.4318,
+      "step": 295
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7136808633804321,
+      "learning_rate": 8.444834595378434e-05,
+      "loss": 0.2916,
+      "step": 296
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8039273619651794,
+      "learning_rate": 8.433689318791626e-05,
+      "loss": 0.3958,
+      "step": 297
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8413447141647339,
+      "learning_rate": 8.422511659797279e-05,
+      "loss": 0.3412,
+      "step": 298
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8425654172897339,
+      "learning_rate": 8.411301723809563e-05,
+      "loss": 0.3274,
+      "step": 299
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8321859836578369,
+      "learning_rate": 8.400059616547047e-05,
+      "loss": 0.3756,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.3578859567642212,
+      "eval_runtime": 359.6405,
+      "eval_samples_per_second": 0.948,
+      "eval_steps_per_second": 0.239,
+      "step": 300
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9729025363922119,
+      "learning_rate": 8.388785444031695e-05,
+      "loss": 0.4602,
+      "step": 301
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7305243611335754,
+      "learning_rate": 8.377479312587879e-05,
+      "loss": 0.3099,
+      "step": 302
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8190518617630005,
+      "learning_rate": 8.366141328841366e-05,
+      "loss": 0.378,
+      "step": 303
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6255936622619629,
+      "learning_rate": 8.354771599718312e-05,
+      "loss": 0.2683,
+      "step": 304
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8600583672523499,
+      "learning_rate": 8.343370232444261e-05,
+      "loss": 0.3999,
+      "step": 305
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7865129113197327,
+      "learning_rate": 8.331937334543132e-05,
+      "loss": 0.3573,
+      "step": 306
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9450405836105347,
+      "learning_rate": 8.320473013836196e-05,
+      "loss": 0.4611,
+      "step": 307
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9604294896125793,
+      "learning_rate": 8.308977378441072e-05,
+      "loss": 0.5008,
+      "step": 308
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.7581052780151367,
+      "learning_rate": 8.297450536770697e-05,
+      "loss": 0.3265,
+      "step": 309
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.865536630153656,
+      "learning_rate": 8.28589259753231e-05,
+      "loss": 0.3223,
+      "step": 310
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8735114336013794,
+      "learning_rate": 8.274303669726426e-05,
+      "loss": 0.4048,
+      "step": 311
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7962080240249634,
+      "learning_rate": 8.262683862645804e-05,
+      "loss": 0.3846,
+      "step": 312
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8916413187980652,
+      "learning_rate": 8.251033285874421e-05,
+      "loss": 0.3826,
+      "step": 313
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9847537279129028,
+      "learning_rate": 8.239352049286435e-05,
+      "loss": 0.3722,
+      "step": 314
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8299481272697449,
+      "learning_rate": 8.22764026304515e-05,
+      "loss": 0.3687,
+      "step": 315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.889231264591217,
+      "learning_rate": 8.21589803760198e-05,
+      "loss": 0.3886,
+      "step": 316
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7379969954490662,
+      "learning_rate": 8.204125483695403e-05,
+      "loss": 0.354,
+      "step": 317
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7494072318077087,
+      "learning_rate": 8.192322712349917e-05,
+      "loss": 0.3039,
+      "step": 318
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8588222861289978,
+      "learning_rate": 8.180489834875e-05,
+      "loss": 0.3703,
+      "step": 319
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7427577376365662,
+      "learning_rate": 8.168626962864045e-05,
+      "loss": 0.2942,
+      "step": 320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9632477760314941,
+      "learning_rate": 8.156734208193327e-05,
+      "loss": 0.368,
+      "step": 321
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7752010226249695,
+      "learning_rate": 8.144811683020932e-05,
+      "loss": 0.3401,
+      "step": 322
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7899407744407654,
+      "learning_rate": 8.132859499785707e-05,
+      "loss": 0.3992,
+      "step": 323
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7520030736923218,
+      "learning_rate": 8.120877771206201e-05,
+      "loss": 0.3631,
+      "step": 324
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.9725971221923828,
+      "learning_rate": 8.108866610279593e-05,
+      "loss": 0.4224,
+      "step": 325
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8367748260498047,
+      "learning_rate": 8.096826130280639e-05,
+      "loss": 0.3653,
+      "step": 326
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8221279978752136,
+      "learning_rate": 8.08475644476059e-05,
+      "loss": 0.3644,
+      "step": 327
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7229669094085693,
+      "learning_rate": 8.072657667546136e-05,
+      "loss": 0.4082,
+      "step": 328
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8275790214538574,
+      "learning_rate": 8.060529912738315e-05,
+      "loss": 0.3986,
+      "step": 329
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8063262104988098,
+      "learning_rate": 8.048373294711455e-05,
+      "loss": 0.3812,
+      "step": 330
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.9632834792137146,
+      "learning_rate": 8.036187928112087e-05,
+      "loss": 0.3777,
+      "step": 331
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8784337043762207,
+      "learning_rate": 8.023973927857857e-05,
+      "loss": 0.448,
+      "step": 332
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9204076528549194,
+      "learning_rate": 8.011731409136454e-05,
+      "loss": 0.3942,
+      "step": 333
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7411671280860901,
+      "learning_rate": 7.99946048740452e-05,
+      "loss": 0.2826,
+      "step": 334
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9016052484512329,
+      "learning_rate": 7.987161278386554e-05,
+      "loss": 0.4256,
+      "step": 335
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9503585696220398,
+      "learning_rate": 7.974833898073832e-05,
+      "loss": 0.3903,
+      "step": 336
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8996967673301697,
+      "learning_rate": 7.962478462723307e-05,
+      "loss": 0.4083,
+      "step": 337
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8242809772491455,
+      "learning_rate": 7.950095088856508e-05,
+      "loss": 0.4013,
+      "step": 338
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7394823431968689,
+      "learning_rate": 7.937683893258454e-05,
+      "loss": 0.2642,
+      "step": 339
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8354219794273376,
+      "learning_rate": 7.925244992976538e-05,
+      "loss": 0.4575,
+      "step": 340
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.738464891910553,
+      "learning_rate": 7.912778505319436e-05,
+      "loss": 0.3643,
+      "step": 341
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6939475536346436,
+      "learning_rate": 7.900284547855991e-05,
+      "loss": 0.2515,
+      "step": 342
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8593461513519287,
+      "learning_rate": 7.88776323841411e-05,
+      "loss": 0.4039,
+      "step": 343
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.899161696434021,
+      "learning_rate": 7.875214695079647e-05,
+      "loss": 0.4574,
+      "step": 344
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.582124650478363,
+      "learning_rate": 7.862639036195298e-05,
+      "loss": 0.2741,
+      "step": 345
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7706286907196045,
+      "learning_rate": 7.850036380359478e-05,
+      "loss": 0.3928,
+      "step": 346
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8360547423362732,
+      "learning_rate": 7.837406846425204e-05,
+      "loss": 0.4336,
+      "step": 347
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8640238046646118,
+      "learning_rate": 7.824750553498976e-05,
+      "loss": 0.4287,
+      "step": 348
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8950493335723877,
+      "learning_rate": 7.812067620939652e-05,
+      "loss": 0.4008,
+      "step": 349
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7541325092315674,
+      "learning_rate": 7.799358168357323e-05,
+      "loss": 0.3444,
+      "step": 350
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8390917778015137,
+      "learning_rate": 7.786622315612183e-05,
+      "loss": 0.4501,
+      "step": 351
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7759553790092468,
+      "learning_rate": 7.773860182813403e-05,
+      "loss": 0.3155,
+      "step": 352
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8678932785987854,
+      "learning_rate": 7.761071890317994e-05,
+      "loss": 0.379,
+      "step": 353
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7860472202301025,
+      "learning_rate": 7.748257558729677e-05,
+      "loss": 0.3407,
+      "step": 354
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7715388536453247,
+      "learning_rate": 7.735417308897736e-05,
+      "loss": 0.4075,
+      "step": 355
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.770388662815094,
+      "learning_rate": 7.72255126191589e-05,
+      "loss": 0.3363,
+      "step": 356
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9575510621070862,
+      "learning_rate": 7.709659539121144e-05,
+      "loss": 0.3703,
+      "step": 357
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7390890717506409,
+      "learning_rate": 7.696742262092644e-05,
+      "loss": 0.3558,
+      "step": 358
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.788749635219574,
+      "learning_rate": 7.683799552650534e-05,
+      "loss": 0.2896,
+      "step": 359
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7049228549003601,
+      "learning_rate": 7.67083153285481e-05,
+      "loss": 0.256,
+      "step": 360
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.72030109167099,
+      "learning_rate": 7.657838325004159e-05,
+      "loss": 0.3029,
+      "step": 361
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8016324639320374,
+      "learning_rate": 7.644820051634812e-05,
+      "loss": 0.3714,
+      "step": 362
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9182860255241394,
+      "learning_rate": 7.63177683551939e-05,
+      "loss": 0.3437,
+      "step": 363
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8547963500022888,
+      "learning_rate": 7.618708799665744e-05,
+      "loss": 0.3474,
+      "step": 364
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9695311188697815,
+      "learning_rate": 7.605616067315793e-05,
+      "loss": 0.4291,
+      "step": 365
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8387336134910583,
+      "learning_rate": 7.592498761944362e-05,
+      "loss": 0.267,
+      "step": 366
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6682276725769043,
+      "learning_rate": 7.579357007258022e-05,
+      "loss": 0.205,
+      "step": 367
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7591363191604614,
+      "learning_rate": 7.56619092719392e-05,
+      "loss": 0.2887,
+      "step": 368
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7149881720542908,
+      "learning_rate": 7.553000645918611e-05,
+      "loss": 0.241,
+      "step": 369
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6557583808898926,
+      "learning_rate": 7.539786287826885e-05,
+      "loss": 0.2455,
+      "step": 370
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6746556162834167,
+      "learning_rate": 7.526547977540592e-05,
+      "loss": 0.2404,
+      "step": 371
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8298879861831665,
+      "learning_rate": 7.51328583990748e-05,
+      "loss": 0.317,
+      "step": 372
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9162379503250122,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.3244,
+      "step": 373
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7807391881942749,
+      "learning_rate": 7.486690583114136e-05,
+      "loss": 0.2825,
+      "step": 374
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9411114454269409,
+      "learning_rate": 7.473357714768222e-05,
+      "loss": 0.3896,
+      "step": 375
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0387256145477295,
+      "learning_rate": 7.460001520701755e-05,
+      "loss": 0.5066,
+      "step": 376
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9648047089576721,
+      "learning_rate": 7.446622126874218e-05,
+      "loss": 0.4065,
+      "step": 377
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9833998084068298,
+      "learning_rate": 7.433219659463881e-05,
+      "loss": 0.421,
+      "step": 378
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9472837448120117,
+      "learning_rate": 7.41979424486662e-05,
+      "loss": 0.4402,
+      "step": 379
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9002391695976257,
+      "learning_rate": 7.406346009694713e-05,
+      "loss": 0.3266,
+      "step": 380
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7711237668991089,
+      "learning_rate": 7.392875080775665e-05,
+      "loss": 0.306,
+      "step": 381
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9758237600326538,
+      "learning_rate": 7.379381585150996e-05,
+      "loss": 0.373,
+      "step": 382
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7764616012573242,
+      "learning_rate": 7.365865650075046e-05,
+      "loss": 0.3121,
+      "step": 383
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7114037275314331,
+      "learning_rate": 7.35232740301378e-05,
+      "loss": 0.2634,
+      "step": 384
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8320745825767517,
+      "learning_rate": 7.338766971643579e-05,
+      "loss": 0.3716,
+      "step": 385
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8671140074729919,
+      "learning_rate": 7.325184483850042e-05,
+      "loss": 0.3343,
+      "step": 386
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6686167120933533,
+      "learning_rate": 7.311580067726783e-05,
+      "loss": 0.2591,
+      "step": 387
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8251988291740417,
+      "learning_rate": 7.297953851574206e-05,
+      "loss": 0.3415,
+      "step": 388
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9778605699539185,
+      "learning_rate": 7.284305963898314e-05,
+      "loss": 0.4871,
+      "step": 389
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.929358959197998,
+      "learning_rate": 7.27063653340949e-05,
+      "loss": 0.3871,
+      "step": 390
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8850415349006653,
+      "learning_rate": 7.25694568902128e-05,
+      "loss": 0.3849,
+      "step": 391
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7394090890884399,
+      "learning_rate": 7.243233559849179e-05,
+      "loss": 0.248,
+      "step": 392
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7275423407554626,
+      "learning_rate": 7.229500275209418e-05,
+      "loss": 0.3003,
+      "step": 393
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8129082322120667,
+      "learning_rate": 7.215745964617737e-05,
+      "loss": 0.3217,
+      "step": 394
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8276209831237793,
+      "learning_rate": 7.201970757788172e-05,
+      "loss": 0.3954,
+      "step": 395
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6335451006889343,
+      "learning_rate": 7.188174784631824e-05,
+      "loss": 0.2639,
+      "step": 396
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7887995839118958,
+      "learning_rate": 7.174358175255636e-05,
+      "loss": 0.3491,
+      "step": 397
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8674720525741577,
+      "learning_rate": 7.160521059961169e-05,
+      "loss": 0.3504,
+      "step": 398
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6642595529556274,
+      "learning_rate": 7.146663569243371e-05,
+      "loss": 0.2619,
+      "step": 399
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9217017889022827,
+      "learning_rate": 7.132785833789344e-05,
+      "loss": 0.4914,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.3484853208065033,
+      "eval_runtime": 403.8922,
+      "eval_samples_per_second": 0.844,
+      "eval_steps_per_second": 0.213,
+      "step": 400
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7344703078269958,
+      "learning_rate": 7.118887984477116e-05,
+      "loss": 0.3356,
+      "step": 401
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7555281519889832,
+      "learning_rate": 7.104970152374405e-05,
+      "loss": 0.3414,
+      "step": 402
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.827525794506073,
+      "learning_rate": 7.091032468737381e-05,
+      "loss": 0.3404,
+      "step": 403
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.0348906517028809,
+      "learning_rate": 7.077075065009433e-05,
+      "loss": 0.3371,
+      "step": 404
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.797475278377533,
+      "learning_rate": 7.063098072819919e-05,
+      "loss": 0.3513,
+      "step": 405
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6458059549331665,
+      "learning_rate": 7.049101623982937e-05,
+      "loss": 0.2531,
+      "step": 406
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6900717616081238,
+      "learning_rate": 7.03508585049608e-05,
+      "loss": 0.2876,
+      "step": 407
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8747743964195251,
+      "learning_rate": 7.021050884539177e-05,
+      "loss": 0.3678,
+      "step": 408
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7450495362281799,
+      "learning_rate": 7.006996858473069e-05,
+      "loss": 0.3026,
+      "step": 409
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7571479678153992,
+      "learning_rate": 6.99292390483834e-05,
+      "loss": 0.2828,
+      "step": 410
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7022895812988281,
+      "learning_rate": 6.97883215635408e-05,
+      "loss": 0.3141,
+      "step": 411
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9226679801940918,
+      "learning_rate": 6.964721745916631e-05,
+      "loss": 0.3961,
+      "step": 412
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8489812016487122,
+      "learning_rate": 6.950592806598327e-05,
+      "loss": 0.3527,
+      "step": 413
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7591996788978577,
+      "learning_rate": 6.936445471646249e-05,
+      "loss": 0.3019,
+      "step": 414
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7016785740852356,
+      "learning_rate": 6.922279874480959e-05,
+      "loss": 0.3708,
+      "step": 415
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7384409308433533,
+      "learning_rate": 6.908096148695251e-05,
+      "loss": 0.3486,
+      "step": 416
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7973634004592896,
+      "learning_rate": 6.89389442805288e-05,
+      "loss": 0.3138,
+      "step": 417
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.160613775253296,
+      "learning_rate": 6.879674846487314e-05,
+      "loss": 0.3139,
+      "step": 418
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8184626698493958,
+      "learning_rate": 6.865437538100457e-05,
+      "loss": 0.3784,
+      "step": 419
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7294750809669495,
+      "learning_rate": 6.851182637161396e-05,
+      "loss": 0.268,
+      "step": 420
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8555173277854919,
+      "learning_rate": 6.836910278105124e-05,
+      "loss": 0.3282,
+      "step": 421
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8913937211036682,
+      "learning_rate": 6.822620595531286e-05,
+      "loss": 0.3871,
+      "step": 422
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8056496977806091,
+      "learning_rate": 6.808313724202894e-05,
+      "loss": 0.3166,
+      "step": 423
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6837203502655029,
+      "learning_rate": 6.793989799045066e-05,
+      "loss": 0.2724,
+      "step": 424
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.944281280040741,
+      "learning_rate": 6.779648955143754e-05,
+      "loss": 0.3812,
+      "step": 425
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8449427485466003,
+      "learning_rate": 6.765291327744463e-05,
+      "loss": 0.3573,
+      "step": 426
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.784675121307373,
+      "learning_rate": 6.750917052250982e-05,
+      "loss": 0.3395,
+      "step": 427
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8955050110816956,
+      "learning_rate": 6.736526264224101e-05,
+      "loss": 0.3777,
+      "step": 428
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7397021651268005,
+      "learning_rate": 6.722119099380345e-05,
+      "loss": 0.3696,
+      "step": 429
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.790101945400238,
+      "learning_rate": 6.707695693590676e-05,
+      "loss": 0.2715,
+      "step": 430
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.7076525688171387,
+      "learning_rate": 6.693256182879225e-05,
+      "loss": 0.3317,
+      "step": 431
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8293165564537048,
+      "learning_rate": 6.678800703422003e-05,
+      "loss": 0.4389,
+      "step": 432
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9080016613006592,
+      "learning_rate": 6.664329391545626e-05,
+      "loss": 0.4216,
+      "step": 433
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8625812530517578,
+      "learning_rate": 6.64984238372601e-05,
+      "loss": 0.4494,
+      "step": 434
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6917742490768433,
+      "learning_rate": 6.635339816587109e-05,
+      "loss": 0.3495,
+      "step": 435
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.805234432220459,
+      "learning_rate": 6.620821826899606e-05,
+      "loss": 0.422,
+      "step": 436
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8176122307777405,
+      "learning_rate": 6.606288551579629e-05,
+      "loss": 0.4044,
+      "step": 437
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.849812924861908,
+      "learning_rate": 6.59174012768747e-05,
+      "loss": 0.442,
+      "step": 438
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7694799900054932,
+      "learning_rate": 6.577176692426279e-05,
+      "loss": 0.3992,
+      "step": 439
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7735298871994019,
+      "learning_rate": 6.562598383140772e-05,
+      "loss": 0.3822,
+      "step": 440
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7290200591087341,
+      "learning_rate": 6.548005337315944e-05,
+      "loss": 0.3631,
+      "step": 441
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5894444584846497,
+      "learning_rate": 6.533397692575766e-05,
+      "loss": 0.2209,
+      "step": 442
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8539714813232422,
+      "learning_rate": 6.518775586681887e-05,
+      "loss": 0.4914,
+      "step": 443
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6574975848197937,
+      "learning_rate": 6.504139157532337e-05,
+      "loss": 0.3166,
+      "step": 444
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7064083814620972,
+      "learning_rate": 6.489488543160224e-05,
+      "loss": 0.2519,
+      "step": 445
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7688191533088684,
+      "learning_rate": 6.47482388173244e-05,
+      "loss": 0.2899,
+      "step": 446
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8815706968307495,
+      "learning_rate": 6.460145311548341e-05,
+      "loss": 0.4429,
+      "step": 447
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8230547308921814,
+      "learning_rate": 6.445452971038464e-05,
+      "loss": 0.3573,
+      "step": 448
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7306635975837708,
+      "learning_rate": 6.430746998763204e-05,
+      "loss": 0.3779,
+      "step": 449
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8090017437934875,
+      "learning_rate": 6.416027533411519e-05,
+      "loss": 0.385,
+      "step": 450
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.0312097072601318,
+      "learning_rate": 6.401294713799618e-05,
+      "loss": 0.4829,
+      "step": 451
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8374634385108948,
+      "learning_rate": 6.386548678869644e-05,
+      "loss": 0.3623,
+      "step": 452
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6865454316139221,
+      "learning_rate": 6.37178956768838e-05,
+      "loss": 0.3109,
+      "step": 453
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7207121849060059,
+      "learning_rate": 6.35701751944592e-05,
+      "loss": 0.3488,
+      "step": 454
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7998125553131104,
+      "learning_rate": 6.342232673454371e-05,
+      "loss": 0.379,
+      "step": 455
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8461378216743469,
+      "learning_rate": 6.32743516914653e-05,
+      "loss": 0.2667,
+      "step": 456
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8442595601081848,
+      "learning_rate": 6.312625146074575e-05,
+      "loss": 0.3888,
+      "step": 457
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7580370903015137,
+      "learning_rate": 6.297802743908741e-05,
+      "loss": 0.3316,
+      "step": 458
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7984005212783813,
+      "learning_rate": 6.282968102436009e-05,
+      "loss": 0.3494,
+      "step": 459
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7054756879806519,
+      "learning_rate": 6.268121361558792e-05,
+      "loss": 0.2775,
+      "step": 460
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7689543962478638,
+      "learning_rate": 6.253262661293604e-05,
+      "loss": 0.3754,
+      "step": 461
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4822267293930054,
+      "learning_rate": 6.238392141769743e-05,
+      "loss": 0.2844,
+      "step": 462
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9492970705032349,
+      "learning_rate": 6.22350994322798e-05,
+      "loss": 0.4449,
+      "step": 463
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7602095007896423,
+      "learning_rate": 6.208616206019224e-05,
+      "loss": 0.3017,
+      "step": 464
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7714800834655762,
+      "learning_rate": 6.193711070603202e-05,
+      "loss": 0.3081,
+      "step": 465
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.687843382358551,
+      "learning_rate": 6.178794677547137e-05,
+      "loss": 0.3176,
+      "step": 466
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8226491212844849,
+      "learning_rate": 6.163867167524419e-05,
+      "loss": 0.3601,
+      "step": 467
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.1454241275787354,
+      "learning_rate": 6.14892868131328e-05,
+      "loss": 0.3769,
+      "step": 468
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8683786392211914,
+      "learning_rate": 6.133979359795468e-05,
+      "loss": 0.3173,
+      "step": 469
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7410722970962524,
+      "learning_rate": 6.119019343954914e-05,
+      "loss": 0.3294,
+      "step": 470
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6974722743034363,
+      "learning_rate": 6.104048774876406e-05,
+      "loss": 0.3424,
+      "step": 471
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8097363114356995,
+      "learning_rate": 6.0890677937442574e-05,
+      "loss": 0.4177,
+      "step": 472
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.989449143409729,
+      "learning_rate": 6.074076541840977e-05,
+      "loss": 0.5301,
+      "step": 473
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.685943067073822,
+      "learning_rate": 6.059075160545933e-05,
+      "loss": 0.2655,
+      "step": 474
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7464625239372253,
+      "learning_rate": 6.0440637913340224e-05,
+      "loss": 0.2567,
+      "step": 475
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7387152314186096,
+      "learning_rate": 6.029042575774334e-05,
+      "loss": 0.3204,
+      "step": 476
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8280574083328247,
+      "learning_rate": 6.0140116555288204e-05,
+      "loss": 0.381,
+      "step": 477
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7447198629379272,
+      "learning_rate": 5.998971172350953e-05,
+      "loss": 0.3358,
+      "step": 478
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6286566853523254,
+      "learning_rate": 5.9839212680843925e-05,
+      "loss": 0.2741,
+      "step": 479
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.7539522051811218,
+      "learning_rate": 5.968862084661643e-05,
+      "loss": 0.4041,
+      "step": 480
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7624731659889221,
+      "learning_rate": 5.953793764102722e-05,
+      "loss": 0.3458,
+      "step": 481
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8316192030906677,
+      "learning_rate": 5.938716448513818e-05,
+      "loss": 0.3268,
+      "step": 482
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8744710087776184,
+      "learning_rate": 5.923630280085948e-05,
+      "loss": 0.491,
+      "step": 483
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9695024490356445,
+      "learning_rate": 5.908535401093618e-05,
+      "loss": 0.4278,
+      "step": 484
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7030946016311646,
+      "learning_rate": 5.8934319538934824e-05,
+      "loss": 0.3356,
+      "step": 485
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6999789476394653,
+      "learning_rate": 5.878320080923001e-05,
+      "loss": 0.3297,
+      "step": 486
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7411117553710938,
+      "learning_rate": 5.863199924699095e-05,
+      "loss": 0.329,
+      "step": 487
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9145691990852356,
+      "learning_rate": 5.848071627816803e-05,
+      "loss": 0.4701,
+      "step": 488
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8818939328193665,
+      "learning_rate": 5.8329353329479366e-05,
+      "loss": 0.4136,
+      "step": 489
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7618187069892883,
+      "learning_rate": 5.8177911828397336e-05,
+      "loss": 0.3021,
+      "step": 490
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8602275848388672,
+      "learning_rate": 5.802639320313514e-05,
+      "loss": 0.4228,
+      "step": 491
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8414841294288635,
+      "learning_rate": 5.787479888263333e-05,
+      "loss": 0.3786,
+      "step": 492
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.9108616709709167,
+      "learning_rate": 5.77231302965463e-05,
+      "loss": 0.3621,
+      "step": 493
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.868047833442688,
+      "learning_rate": 5.757138887522884e-05,
+      "loss": 0.3343,
+      "step": 494
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6807160377502441,
+      "learning_rate": 5.7419576049722646e-05,
+      "loss": 0.2953,
+      "step": 495
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7385652661323547,
+      "learning_rate": 5.7267693251742794e-05,
+      "loss": 0.3439,
+      "step": 496
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8206238746643066,
+      "learning_rate": 5.7115741913664264e-05,
+      "loss": 0.376,
+      "step": 497
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8396340608596802,
+      "learning_rate": 5.696372346850841e-05,
+      "loss": 0.4251,
+      "step": 498
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7308201789855957,
+      "learning_rate": 5.68116393499295e-05,
+      "loss": 0.3765,
+      "step": 499
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6886759996414185,
+      "learning_rate": 5.6659490992201094e-05,
+      "loss": 0.2597,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.3377499282360077,
+      "eval_runtime": 397.2522,
+      "eval_samples_per_second": 0.858,
+      "eval_steps_per_second": 0.216,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8282071352005005,
+      "learning_rate": 5.650727983020262e-05,
+      "loss": 0.5015,
+      "step": 501
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6353948712348938,
+      "learning_rate": 5.635500729940577e-05,
+      "loss": 0.2603,
+      "step": 502
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.898929238319397,
+      "learning_rate": 5.620267483586105e-05,
+      "loss": 0.4508,
+      "step": 503
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7678223252296448,
+      "learning_rate": 5.6050283876184116e-05,
+      "loss": 0.4265,
+      "step": 504
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.006208896636963,
+      "learning_rate": 5.5897835857542317e-05,
+      "loss": 0.4317,
+      "step": 505
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7651686668395996,
+      "learning_rate": 5.574533221764109e-05,
+      "loss": 0.2968,
+      "step": 506
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6841700077056885,
+      "learning_rate": 5.5592774394710465e-05,
+      "loss": 0.3343,
+      "step": 507
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7523791790008545,
+      "learning_rate": 5.5440163827491454e-05,
+      "loss": 0.3144,
+      "step": 508
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7760979533195496,
+      "learning_rate": 5.5287501955222444e-05,
+      "loss": 0.3503,
+      "step": 509
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7066985964775085,
+      "learning_rate": 5.513479021762572e-05,
+      "loss": 0.3301,
+      "step": 510
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7595872282981873,
+      "learning_rate": 5.4982030054893785e-05,
+      "loss": 0.2965,
+      "step": 511
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.5664269924163818,
+      "learning_rate": 5.482922290767589e-05,
+      "loss": 0.3208,
+      "step": 512
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7117363214492798,
+      "learning_rate": 5.467637021706438e-05,
+      "loss": 0.3303,
+      "step": 513
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8676652908325195,
+      "learning_rate": 5.452347342458104e-05,
+      "loss": 0.4647,
+      "step": 514
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8572256565093994,
+      "learning_rate": 5.437053397216364e-05,
+      "loss": 0.3743,
+      "step": 515
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8073871731758118,
+      "learning_rate": 5.4217553302152237e-05,
+      "loss": 0.3632,
+      "step": 516
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.9069568514823914,
+      "learning_rate": 5.406453285727564e-05,
+      "loss": 0.4266,
+      "step": 517
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7989234924316406,
+      "learning_rate": 5.39114740806377e-05,
+      "loss": 0.3284,
+      "step": 518
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5894029140472412,
+      "learning_rate": 5.3758378415703825e-05,
+      "loss": 0.2744,
+      "step": 519
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.9477267265319824,
+      "learning_rate": 5.360524730628728e-05,
+      "loss": 0.464,
+      "step": 520
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7913030982017517,
+      "learning_rate": 5.345208219653561e-05,
+      "loss": 0.3249,
+      "step": 521
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7727319598197937,
+      "learning_rate": 5.3298884530917014e-05,
+      "loss": 0.3476,
+      "step": 522
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.682256817817688,
+      "learning_rate": 5.3145655754206714e-05,
+      "loss": 0.2899,
+      "step": 523
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5948423743247986,
+      "learning_rate": 5.2992397311473316e-05,
+      "loss": 0.2362,
+      "step": 524
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8064485788345337,
+      "learning_rate": 5.2839110648065214e-05,
+      "loss": 0.3843,
+      "step": 525
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.9509373903274536,
+      "learning_rate": 5.2685797209596976e-05,
+      "loss": 0.3841,
+      "step": 526
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8427961468696594,
+      "learning_rate": 5.2532458441935636e-05,
+      "loss": 0.3352,
+      "step": 527
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7945504784584045,
+      "learning_rate": 5.2379095791187124e-05,
+      "loss": 0.3848,
+      "step": 528
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7354018092155457,
+      "learning_rate": 5.222571070368257e-05,
+      "loss": 0.3412,
+      "step": 529
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.7190101146697998,
+      "learning_rate": 5.2072304625964785e-05,
+      "loss": 0.3351,
+      "step": 530
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.780085027217865,
+      "learning_rate": 5.191887900477443e-05,
+      "loss": 0.3959,
+      "step": 531
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8177609443664551,
+      "learning_rate": 5.176543528703657e-05,
+      "loss": 0.3789,
+      "step": 532
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7294226288795471,
+      "learning_rate": 5.161197491984684e-05,
+      "loss": 0.3028,
+      "step": 533
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8375429511070251,
+      "learning_rate": 5.1458499350458e-05,
+      "loss": 0.2667,
+      "step": 534
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.9048738479614258,
+      "learning_rate": 5.130501002626609e-05,
+      "loss": 0.5359,
+      "step": 535
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7356242537498474,
+      "learning_rate": 5.11515083947969e-05,
+      "loss": 0.3143,
+      "step": 536
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.793245255947113,
+      "learning_rate": 5.099799590369231e-05,
+      "loss": 0.3689,
+      "step": 537
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.840069591999054,
+      "learning_rate": 5.084447400069655e-05,
+      "loss": 0.4039,
+      "step": 538
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7241114377975464,
+      "learning_rate": 5.069094413364272e-05,
+      "loss": 0.3503,
+      "step": 539
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7574796676635742,
+      "learning_rate": 5.053740775043892e-05,
+      "loss": 0.385,
+      "step": 540
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7796861529350281,
+      "learning_rate": 5.038386629905475e-05,
+      "loss": 0.3863,
+      "step": 541
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8066446781158447,
+      "learning_rate": 5.023032122750759e-05,
+      "loss": 0.3786,
+      "step": 542
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.7452917695045471,
+      "learning_rate": 5.007677398384901e-05,
+      "loss": 0.31,
+      "step": 543
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7428964972496033,
+      "learning_rate": 4.9923226016151e-05,
+      "loss": 0.4059,
+      "step": 544
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7531483769416809,
+      "learning_rate": 4.976967877249242e-05,
+      "loss": 0.3843,
+      "step": 545
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6530309915542603,
+      "learning_rate": 4.961613370094527e-05,
+      "loss": 0.2778,
+      "step": 546
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7551013231277466,
+      "learning_rate": 4.9462592249561095e-05,
+      "loss": 0.4133,
+      "step": 547
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6585403680801392,
+      "learning_rate": 4.9309055866357295e-05,
+      "loss": 0.2624,
+      "step": 548
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7526929974555969,
+      "learning_rate": 4.915552599930345e-05,
+      "loss": 0.4082,
+      "step": 549
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7001590132713318,
+      "learning_rate": 4.900200409630771e-05,
+      "loss": 0.3205,
+      "step": 550
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6295161247253418,
+      "learning_rate": 4.884849160520311e-05,
+      "loss": 0.2718,
+      "step": 551
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7315431237220764,
+      "learning_rate": 4.869498997373393e-05,
+      "loss": 0.3426,
+      "step": 552
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8753539323806763,
+      "learning_rate": 4.8541500649542014e-05,
+      "loss": 0.4409,
+      "step": 553
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7550821304321289,
+      "learning_rate": 4.838802508015316e-05,
+      "loss": 0.3561,
+      "step": 554
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.156670093536377,
+      "learning_rate": 4.823456471296344e-05,
+      "loss": 0.3641,
+      "step": 555
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7014812231063843,
+      "learning_rate": 4.8081120995225574e-05,
+      "loss": 0.3109,
+      "step": 556
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7548364400863647,
+      "learning_rate": 4.792769537403523e-05,
+      "loss": 0.4466,
+      "step": 557
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8029537796974182,
+      "learning_rate": 4.777428929631743e-05,
+      "loss": 0.4332,
+      "step": 558
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8440172672271729,
+      "learning_rate": 4.762090420881289e-05,
+      "loss": 0.4385,
+      "step": 559
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.922404944896698,
+      "learning_rate": 4.746754155806437e-05,
+      "loss": 0.5124,
+      "step": 560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8025408983230591,
+      "learning_rate": 4.731420279040303e-05,
+      "loss": 0.362,
+      "step": 561
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6226940751075745,
+      "learning_rate": 4.716088935193479e-05,
+      "loss": 0.2139,
+      "step": 562
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.7734771966934204,
+      "learning_rate": 4.7007602688526695e-05,
+      "loss": 0.392,
+      "step": 563
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.581691324710846,
+      "learning_rate": 4.6854344245793305e-05,
+      "loss": 0.2372,
+      "step": 564
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6351912021636963,
+      "learning_rate": 4.6701115469083e-05,
+      "loss": 0.2883,
+      "step": 565
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6465301513671875,
+      "learning_rate": 4.65479178034644e-05,
+      "loss": 0.2229,
+      "step": 566
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7297216057777405,
+      "learning_rate": 4.6394752693712734e-05,
+      "loss": 0.3287,
+      "step": 567
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8667015433311462,
+      "learning_rate": 4.624162158429619e-05,
+      "loss": 0.3055,
+      "step": 568
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9637848138809204,
+      "learning_rate": 4.608852591936231e-05,
+      "loss": 0.4632,
+      "step": 569
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.3064755201339722,
+      "learning_rate": 4.593546714272438e-05,
+      "loss": 0.3813,
+      "step": 570
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8329745531082153,
+      "learning_rate": 4.578244669784777e-05,
+      "loss": 0.3986,
+      "step": 571
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5878795981407166,
+      "learning_rate": 4.562946602783636e-05,
+      "loss": 0.2159,
+      "step": 572
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9630850553512573,
+      "learning_rate": 4.547652657541897e-05,
+      "loss": 0.5043,
+      "step": 573
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8134487271308899,
+      "learning_rate": 4.532362978293564e-05,
+      "loss": 0.4179,
+      "step": 574
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7672256827354431,
+      "learning_rate": 4.517077709232411e-05,
+      "loss": 0.2967,
+      "step": 575
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.897541880607605,
+      "learning_rate": 4.501796994510622e-05,
+      "loss": 0.4467,
+      "step": 576
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5795857310295105,
+      "learning_rate": 4.4865209782374304e-05,
+      "loss": 0.2284,
+      "step": 577
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6907414197921753,
+      "learning_rate": 4.471249804477758e-05,
+      "loss": 0.2545,
+      "step": 578
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8034598231315613,
+      "learning_rate": 4.4559836172508565e-05,
+      "loss": 0.3856,
+      "step": 579
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6842982172966003,
+      "learning_rate": 4.4407225605289547e-05,
+      "loss": 0.2685,
+      "step": 580
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6309784054756165,
+      "learning_rate": 4.4254667782358924e-05,
+      "loss": 0.2661,
+      "step": 581
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8462599515914917,
+      "learning_rate": 4.410216414245771e-05,
+      "loss": 0.4423,
+      "step": 582
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7399506568908691,
+      "learning_rate": 4.394971612381591e-05,
+      "loss": 0.3249,
+      "step": 583
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.7370103001594543,
+      "learning_rate": 4.379732516413897e-05,
+      "loss": 0.2999,
+      "step": 584
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.907970666885376,
+      "learning_rate": 4.364499270059424e-05,
+      "loss": 0.3704,
+      "step": 585
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7563538551330566,
+      "learning_rate": 4.34927201697974e-05,
+      "loss": 0.3297,
+      "step": 586
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7526343464851379,
+      "learning_rate": 4.334050900779893e-05,
+      "loss": 0.335,
+      "step": 587
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8954048156738281,
+      "learning_rate": 4.3188360650070525e-05,
+      "loss": 0.4079,
+      "step": 588
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6989839673042297,
+      "learning_rate": 4.303627653149159e-05,
+      "loss": 0.2647,
+      "step": 589
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7470086216926575,
+      "learning_rate": 4.288425808633575e-05,
+      "loss": 0.2902,
+      "step": 590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8542561531066895,
+      "learning_rate": 4.2732306748257224e-05,
+      "loss": 0.3554,
+      "step": 591
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6488617658615112,
+      "learning_rate": 4.258042395027738e-05,
+      "loss": 0.2746,
+      "step": 592
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.9786548018455505,
+      "learning_rate": 4.2428611124771184e-05,
+      "loss": 0.3488,
+      "step": 593
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7049300670623779,
+      "learning_rate": 4.2276869703453724e-05,
+      "loss": 0.3409,
+      "step": 594
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7597625851631165,
+      "learning_rate": 4.21252011173667e-05,
+      "loss": 0.3082,
+      "step": 595
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.8109071850776672,
+      "learning_rate": 4.1973606796864884e-05,
+      "loss": 0.3214,
+      "step": 596
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8080856800079346,
+      "learning_rate": 4.182208817160269e-05,
+      "loss": 0.4177,
+      "step": 597
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6627965569496155,
+      "learning_rate": 4.167064667052065e-05,
+      "loss": 0.2974,
+      "step": 598
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.762102484703064,
+      "learning_rate": 4.151928372183198e-05,
+      "loss": 0.3576,
+      "step": 599
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7814170718193054,
+      "learning_rate": 4.136800075300906e-05,
+      "loss": 0.3593,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.33276209235191345,
+      "eval_runtime": 392.5383,
+      "eval_samples_per_second": 0.869,
+      "eval_steps_per_second": 0.219,
+      "step": 600
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.105913758277893,
+      "learning_rate": 4.1216799190770006e-05,
+      "loss": 0.4906,
+      "step": 601
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.581840455532074,
+      "learning_rate": 4.1065680461065194e-05,
+      "loss": 0.2692,
+      "step": 602
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8193796873092651,
+      "learning_rate": 4.0914645989063845e-05,
+      "loss": 0.3474,
+      "step": 603
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7987298965454102,
+      "learning_rate": 4.076369719914055e-05,
+      "loss": 0.32,
+      "step": 604
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.9855241179466248,
+      "learning_rate": 4.061283551486185e-05,
+      "loss": 0.4974,
+      "step": 605
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.9153417944908142,
+      "learning_rate": 4.0462062358972796e-05,
+      "loss": 0.4414,
+      "step": 606
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.9190502762794495,
+      "learning_rate": 4.03113791533836e-05,
+      "loss": 0.4704,
+      "step": 607
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7812002301216125,
+      "learning_rate": 4.016078731915608e-05,
+      "loss": 0.3457,
+      "step": 608
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7470685243606567,
+      "learning_rate": 4.001028827649046e-05,
+      "loss": 0.3595,
+      "step": 609
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.715591311454773,
+      "learning_rate": 3.9859883444711794e-05,
+      "loss": 0.294,
+      "step": 610
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6239838004112244,
+      "learning_rate": 3.970957424225666e-05,
+      "loss": 0.2688,
+      "step": 611
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7052245140075684,
+      "learning_rate": 3.955936208665979e-05,
+      "loss": 0.2874,
+      "step": 612
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7291432023048401,
+      "learning_rate": 3.940924839454066e-05,
+      "loss": 0.3193,
+      "step": 613
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8084625005722046,
+      "learning_rate": 3.925923458159023e-05,
+      "loss": 0.4363,
+      "step": 614
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7001256942749023,
+      "learning_rate": 3.9109322062557424e-05,
+      "loss": 0.3071,
+      "step": 615
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6469643712043762,
+      "learning_rate": 3.895951225123595e-05,
+      "loss": 0.2536,
+      "step": 616
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7200465202331543,
+      "learning_rate": 3.880980656045087e-05,
+      "loss": 0.3389,
+      "step": 617
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7524146437644958,
+      "learning_rate": 3.866020640204533e-05,
+      "loss": 0.3282,
+      "step": 618
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6847262978553772,
+      "learning_rate": 3.851071318686721e-05,
+      "loss": 0.2957,
+      "step": 619
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.693108081817627,
+      "learning_rate": 3.8361328324755825e-05,
+      "loss": 0.2774,
+      "step": 620
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8072023987770081,
+      "learning_rate": 3.821205322452863e-05,
+      "loss": 0.3945,
+      "step": 621
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8001756072044373,
+      "learning_rate": 3.8062889293967976e-05,
+      "loss": 0.469,
+      "step": 622
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7824945449829102,
+      "learning_rate": 3.791383793980776e-05,
+      "loss": 0.3666,
+      "step": 623
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6339123249053955,
+      "learning_rate": 3.7764900567720196e-05,
+      "loss": 0.2611,
+      "step": 624
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8985951542854309,
+      "learning_rate": 3.761607858230257e-05,
+      "loss": 0.4484,
+      "step": 625
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.8501868844032288,
+      "learning_rate": 3.746737338706397e-05,
+      "loss": 0.2901,
+      "step": 626
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.812579870223999,
+      "learning_rate": 3.7318786384412074e-05,
+      "loss": 0.4177,
+      "step": 627
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6757286787033081,
+      "learning_rate": 3.71703189756399e-05,
+      "loss": 0.3659,
+      "step": 628
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7115470170974731,
+      "learning_rate": 3.70219725609126e-05,
+      "loss": 0.317,
+      "step": 629
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7170284390449524,
+      "learning_rate": 3.6873748539254246e-05,
+      "loss": 0.2874,
+      "step": 630
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8659657835960388,
+      "learning_rate": 3.6725648308534696e-05,
+      "loss": 0.4354,
+      "step": 631
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.9819113612174988,
+      "learning_rate": 3.65776732654563e-05,
+      "loss": 0.3852,
+      "step": 632
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8020215630531311,
+      "learning_rate": 3.6429824805540815e-05,
+      "loss": 0.4141,
+      "step": 633
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6901425123214722,
+      "learning_rate": 3.628210432311621e-05,
+      "loss": 0.2453,
+      "step": 634
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.903791606426239,
+      "learning_rate": 3.6134513211303556e-05,
+      "loss": 0.4269,
+      "step": 635
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8659561276435852,
+      "learning_rate": 3.598705286200382e-05,
+      "loss": 0.3723,
+      "step": 636
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8722596764564514,
+      "learning_rate": 3.58397246658848e-05,
+      "loss": 0.4137,
+      "step": 637
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8540229201316833,
+      "learning_rate": 3.5692530012367955e-05,
+      "loss": 0.3733,
+      "step": 638
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8218057155609131,
+      "learning_rate": 3.554547028961537e-05,
+      "loss": 0.3518,
+      "step": 639
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7018035054206848,
+      "learning_rate": 3.5398546884516604e-05,
+      "loss": 0.2624,
+      "step": 640
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.745732843875885,
+      "learning_rate": 3.5251761182675625e-05,
+      "loss": 0.3663,
+      "step": 641
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.9503482580184937,
+      "learning_rate": 3.510511456839777e-05,
+      "loss": 0.3798,
+      "step": 642
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8680411577224731,
+      "learning_rate": 3.495860842467664e-05,
+      "loss": 0.4054,
+      "step": 643
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7186248898506165,
+      "learning_rate": 3.481224413318114e-05,
+      "loss": 0.3766,
+      "step": 644
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8567414879798889,
+      "learning_rate": 3.4666023074242356e-05,
+      "loss": 0.3834,
+      "step": 645
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8206800222396851,
+      "learning_rate": 3.4519946626840573e-05,
+      "loss": 0.3849,
+      "step": 646
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8528415560722351,
+      "learning_rate": 3.4374016168592296e-05,
+      "loss": 0.3565,
+      "step": 647
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8035612106323242,
+      "learning_rate": 3.422823307573722e-05,
+      "loss": 0.4023,
+      "step": 648
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8103017210960388,
+      "learning_rate": 3.40825987231253e-05,
+      "loss": 0.4347,
+      "step": 649
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7140933275222778,
+      "learning_rate": 3.393711448420372e-05,
+      "loss": 0.3276,
+      "step": 650
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7567194700241089,
+      "learning_rate": 3.379178173100396e-05,
+      "loss": 0.327,
+      "step": 651
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6482555270195007,
+      "learning_rate": 3.364660183412892e-05,
+      "loss": 0.2334,
+      "step": 652
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7907357811927795,
+      "learning_rate": 3.3501576162739904e-05,
+      "loss": 0.3407,
+      "step": 653
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7791637182235718,
+      "learning_rate": 3.335670608454376e-05,
+      "loss": 0.3776,
+      "step": 654
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6720924973487854,
+      "learning_rate": 3.321199296577998e-05,
+      "loss": 0.2397,
+      "step": 655
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8295161724090576,
+      "learning_rate": 3.3067438171207766e-05,
+      "loss": 0.4131,
+      "step": 656
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.8072582483291626,
+      "learning_rate": 3.292304306409325e-05,
+      "loss": 0.3459,
+      "step": 657
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7293255925178528,
+      "learning_rate": 3.2778809006196564e-05,
+      "loss": 0.3381,
+      "step": 658
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6666536331176758,
+      "learning_rate": 3.263473735775899e-05,
+      "loss": 0.2396,
+      "step": 659
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7870211005210876,
+      "learning_rate": 3.249082947749019e-05,
+      "loss": 0.2812,
+      "step": 660
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7173169851303101,
+      "learning_rate": 3.234708672255538e-05,
+      "loss": 0.3403,
+      "step": 661
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6862433552742004,
+      "learning_rate": 3.220351044856247e-05,
+      "loss": 0.2891,
+      "step": 662
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7917356491088867,
+      "learning_rate": 3.206010200954935e-05,
+      "loss": 0.3277,
+      "step": 663
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.89388507604599,
+      "learning_rate": 3.191686275797107e-05,
+      "loss": 0.4419,
+      "step": 664
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7249747514724731,
+      "learning_rate": 3.177379404468715e-05,
+      "loss": 0.3237,
+      "step": 665
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5826414823532104,
+      "learning_rate": 3.163089721894876e-05,
+      "loss": 0.2324,
+      "step": 666
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7463696599006653,
+      "learning_rate": 3.148817362838606e-05,
+      "loss": 0.3422,
+      "step": 667
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.84174644947052,
+      "learning_rate": 3.134562461899545e-05,
+      "loss": 0.3687,
+      "step": 668
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7134294509887695,
+      "learning_rate": 3.120325153512687e-05,
+      "loss": 0.2878,
+      "step": 669
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8264583349227905,
+      "learning_rate": 3.10610557194712e-05,
+      "loss": 0.3171,
+      "step": 670
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8185860514640808,
+      "learning_rate": 3.091903851304751e-05,
+      "loss": 0.4482,
+      "step": 671
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.692940354347229,
+      "learning_rate": 3.077720125519042e-05,
+      "loss": 0.3008,
+      "step": 672
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8749237060546875,
+      "learning_rate": 3.063554528353752e-05,
+      "loss": 0.3827,
+      "step": 673
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.661910355091095,
+      "learning_rate": 3.0494071934016737e-05,
+      "loss": 0.2697,
+      "step": 674
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.4601095914840698,
+      "learning_rate": 3.03527825408337e-05,
+      "loss": 0.3928,
+      "step": 675
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7160587906837463,
+      "learning_rate": 3.0211678436459212e-05,
+      "loss": 0.301,
+      "step": 676
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8605940937995911,
+      "learning_rate": 3.0070760951616618e-05,
+      "loss": 0.3647,
+      "step": 677
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8230734467506409,
+      "learning_rate": 2.993003141526933e-05,
+      "loss": 0.3688,
+      "step": 678
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.685835063457489,
+      "learning_rate": 2.978949115460824e-05,
+      "loss": 0.2429,
+      "step": 679
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7387111186981201,
+      "learning_rate": 2.9649141495039223e-05,
+      "loss": 0.319,
+      "step": 680
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.777783215045929,
+      "learning_rate": 2.950898376017064e-05,
+      "loss": 0.3598,
+      "step": 681
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7240700721740723,
+      "learning_rate": 2.9369019271800823e-05,
+      "loss": 0.2636,
+      "step": 682
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8164922595024109,
+      "learning_rate": 2.9229249349905684e-05,
+      "loss": 0.3948,
+      "step": 683
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7590022087097168,
+      "learning_rate": 2.908967531262618e-05,
+      "loss": 0.2638,
+      "step": 684
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7859776020050049,
+      "learning_rate": 2.895029847625595e-05,
+      "loss": 0.3398,
+      "step": 685
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.030025601387024,
+      "learning_rate": 2.8811120155228844e-05,
+      "loss": 0.4647,
+      "step": 686
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.892189085483551,
+      "learning_rate": 2.8672141662106578e-05,
+      "loss": 0.3618,
+      "step": 687
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9350185990333557,
+      "learning_rate": 2.853336430756631e-05,
+      "loss": 0.3879,
+      "step": 688
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8574723601341248,
+      "learning_rate": 2.8394789400388328e-05,
+      "loss": 0.4345,
+      "step": 689
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8923603296279907,
+      "learning_rate": 2.8256418247443662e-05,
+      "loss": 0.387,
+      "step": 690
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7549290657043457,
+      "learning_rate": 2.811825215368179e-05,
+      "loss": 0.3107,
+      "step": 691
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.0609707832336426,
+      "learning_rate": 2.798029242211828e-05,
+      "loss": 0.4496,
+      "step": 692
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7725462913513184,
+      "learning_rate": 2.7842540353822632e-05,
+      "loss": 0.3094,
+      "step": 693
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8404128551483154,
+      "learning_rate": 2.770499724790584e-05,
+      "loss": 0.3208,
+      "step": 694
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8126529455184937,
+      "learning_rate": 2.7567664401508225e-05,
+      "loss": 0.37,
+      "step": 695
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.9001474976539612,
+      "learning_rate": 2.7430543109787222e-05,
+      "loss": 0.4398,
+      "step": 696
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7322224378585815,
+      "learning_rate": 2.729363466590511e-05,
+      "loss": 0.2661,
+      "step": 697
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.7904149293899536,
+      "learning_rate": 2.7156940361016864e-05,
+      "loss": 0.3227,
+      "step": 698
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8572664856910706,
+      "learning_rate": 2.702046148425795e-05,
+      "loss": 0.3378,
+      "step": 699
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.9129036068916321,
+      "learning_rate": 2.688419932273219e-05,
+      "loss": 0.4495,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.32787904143333435,
+      "eval_runtime": 367.0915,
+      "eval_samples_per_second": 0.929,
+      "eval_steps_per_second": 0.234,
+      "step": 700
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.2831424474716187,
+      "learning_rate": 2.6748155161499567e-05,
+      "loss": 0.4302,
+      "step": 701
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6287684440612793,
+      "learning_rate": 2.6612330283564223e-05,
+      "loss": 0.2414,
+      "step": 702
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6575673222541809,
+      "learning_rate": 2.6476725969862227e-05,
+      "loss": 0.2483,
+      "step": 703
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6362529397010803,
+      "learning_rate": 2.634134349924956e-05,
+      "loss": 0.2731,
+      "step": 704
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7954411506652832,
+      "learning_rate": 2.6206184148490066e-05,
+      "loss": 0.3321,
+      "step": 705
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8282104134559631,
+      "learning_rate": 2.6071249192243363e-05,
+      "loss": 0.3724,
+      "step": 706
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7583096027374268,
+      "learning_rate": 2.5936539903052892e-05,
+      "loss": 0.3634,
+      "step": 707
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.8519490361213684,
+      "learning_rate": 2.580205755133384e-05,
+      "loss": 0.4251,
+      "step": 708
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.789387047290802,
+      "learning_rate": 2.5667803405361213e-05,
+      "loss": 0.4324,
+      "step": 709
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9099074006080627,
+      "learning_rate": 2.5533778731257824e-05,
+      "loss": 0.441,
+      "step": 710
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.752293050289154,
+      "learning_rate": 2.5399984792982457e-05,
+      "loss": 0.3258,
+      "step": 711
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7920708656311035,
+      "learning_rate": 2.5266422852317795e-05,
+      "loss": 0.3339,
+      "step": 712
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8558526635169983,
+      "learning_rate": 2.513309416885865e-05,
+      "loss": 0.4195,
+      "step": 713
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7054266929626465,
+      "learning_rate": 2.500000000000001e-05,
+      "loss": 0.2909,
+      "step": 714
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7802085280418396,
+      "learning_rate": 2.4867141600925214e-05,
+      "loss": 0.3372,
+      "step": 715
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7936347723007202,
+      "learning_rate": 2.4734520224594093e-05,
+      "loss": 0.3999,
+      "step": 716
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8295683264732361,
+      "learning_rate": 2.4602137121731195e-05,
+      "loss": 0.3717,
+      "step": 717
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.7945050001144409,
+      "learning_rate": 2.44699935408139e-05,
+      "loss": 0.349,
+      "step": 718
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5299955010414124,
+      "learning_rate": 2.433809072806081e-05,
+      "loss": 0.1798,
+      "step": 719
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8824771642684937,
+      "learning_rate": 2.4206429927419794e-05,
+      "loss": 0.3225,
+      "step": 720
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.718833863735199,
+      "learning_rate": 2.40750123805564e-05,
+      "loss": 0.3222,
+      "step": 721
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8872383832931519,
+      "learning_rate": 2.3943839326842092e-05,
+      "loss": 0.3984,
+      "step": 722
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7282717227935791,
+      "learning_rate": 2.3812912003342568e-05,
+      "loss": 0.2964,
+      "step": 723
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8031437993049622,
+      "learning_rate": 2.3682231644806108e-05,
+      "loss": 0.369,
+      "step": 724
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7498611211776733,
+      "learning_rate": 2.3551799483651894e-05,
+      "loss": 0.3513,
+      "step": 725
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8940857648849487,
+      "learning_rate": 2.342161674995843e-05,
+      "loss": 0.4307,
+      "step": 726
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.743716835975647,
+      "learning_rate": 2.3291684671451903e-05,
+      "loss": 0.3552,
+      "step": 727
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6891101002693176,
+      "learning_rate": 2.3162004473494657e-05,
+      "loss": 0.2907,
+      "step": 728
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7215185761451721,
+      "learning_rate": 2.3032577379073577e-05,
+      "loss": 0.3241,
+      "step": 729
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6909754872322083,
+      "learning_rate": 2.2903404608788582e-05,
+      "loss": 0.2727,
+      "step": 730
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7908779978752136,
+      "learning_rate": 2.2774487380841115e-05,
+      "loss": 0.3612,
+      "step": 731
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7863151431083679,
+      "learning_rate": 2.2645826911022655e-05,
+      "loss": 0.359,
+      "step": 732
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.7265268564224243,
+      "learning_rate": 2.2517424412703252e-05,
+      "loss": 0.3149,
+      "step": 733
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8164087533950806,
+      "learning_rate": 2.2389281096820075e-05,
+      "loss": 0.29,
+      "step": 734
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8119409084320068,
+      "learning_rate": 2.2261398171865978e-05,
+      "loss": 0.3496,
+      "step": 735
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7295995354652405,
+      "learning_rate": 2.2133776843878186e-05,
+      "loss": 0.2933,
+      "step": 736
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7590300440788269,
+      "learning_rate": 2.2006418316426775e-05,
+      "loss": 0.3186,
+      "step": 737
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7812785506248474,
+      "learning_rate": 2.187932379060348e-05,
+      "loss": 0.3547,
+      "step": 738
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.834797739982605,
+      "learning_rate": 2.1752494465010242e-05,
+      "loss": 0.2738,
+      "step": 739
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.799625813961029,
+      "learning_rate": 2.1625931535747964e-05,
+      "loss": 0.3736,
+      "step": 740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7818556427955627,
+      "learning_rate": 2.1499636196405228e-05,
+      "loss": 0.3785,
+      "step": 741
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7916277647018433,
+      "learning_rate": 2.1373609638047032e-05,
+      "loss": 0.3296,
+      "step": 742
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7689991593360901,
+      "learning_rate": 2.1247853049203543e-05,
+      "loss": 0.3121,
+      "step": 743
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7208607792854309,
+      "learning_rate": 2.112236761585892e-05,
+      "loss": 0.3161,
+      "step": 744
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7483397722244263,
+      "learning_rate": 2.09971545214401e-05,
+      "loss": 0.3481,
+      "step": 745
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8467214703559875,
+      "learning_rate": 2.087221494680563e-05,
+      "loss": 0.3671,
+      "step": 746
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6154919266700745,
+      "learning_rate": 2.074755007023461e-05,
+      "loss": 0.2431,
+      "step": 747
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5545762181282043,
+      "learning_rate": 2.0623161067415463e-05,
+      "loss": 0.1903,
+      "step": 748
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7736625075340271,
+      "learning_rate": 2.049904911143492e-05,
+      "loss": 0.3434,
+      "step": 749
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7429782748222351,
+      "learning_rate": 2.0375215372766944e-05,
+      "loss": 0.3448,
+      "step": 750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7946716547012329,
+      "learning_rate": 2.025166101926168e-05,
+      "loss": 0.4484,
+      "step": 751
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.663598895072937,
+      "learning_rate": 2.012838721613447e-05,
+      "loss": 0.2482,
+      "step": 752
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.771665632724762,
+      "learning_rate": 2.0005395125954812e-05,
+      "loss": 0.413,
+      "step": 753
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.6896969676017761,
+      "learning_rate": 1.9882685908635462e-05,
+      "loss": 0.3139,
+      "step": 754
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8102360367774963,
+      "learning_rate": 1.9760260721421426e-05,
+      "loss": 0.3592,
+      "step": 755
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.787959098815918,
+      "learning_rate": 1.963812071887913e-05,
+      "loss": 0.3842,
+      "step": 756
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.727848470211029,
+      "learning_rate": 1.951626705288544e-05,
+      "loss": 0.3537,
+      "step": 757
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7215611934661865,
+      "learning_rate": 1.9394700872616855e-05,
+      "loss": 0.2549,
+      "step": 758
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.883029043674469,
+      "learning_rate": 1.9273423324538658e-05,
+      "loss": 0.4213,
+      "step": 759
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6387115120887756,
+      "learning_rate": 1.9152435552394104e-05,
+      "loss": 0.2798,
+      "step": 760
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7429985404014587,
+      "learning_rate": 1.903173869719362e-05,
+      "loss": 0.2806,
+      "step": 761
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6558197736740112,
+      "learning_rate": 1.891133389720407e-05,
+      "loss": 0.3061,
+      "step": 762
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.699653685092926,
+      "learning_rate": 1.8791222287937982e-05,
+      "loss": 0.3468,
+      "step": 763
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.732164204120636,
+      "learning_rate": 1.8671405002142918e-05,
+      "loss": 0.2682,
+      "step": 764
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.7670356631278992,
+      "learning_rate": 1.855188316979068e-05,
+      "loss": 0.3591,
+      "step": 765
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7469373345375061,
+      "learning_rate": 1.8432657918066732e-05,
+      "loss": 0.3262,
+      "step": 766
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.821238100528717,
+      "learning_rate": 1.831373037135955e-05,
+      "loss": 0.4206,
+      "step": 767
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6317629814147949,
+      "learning_rate": 1.8195101651250017e-05,
+      "loss": 0.2437,
+      "step": 768
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8316160440444946,
+      "learning_rate": 1.807677287650083e-05,
+      "loss": 0.3305,
+      "step": 769
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.686754584312439,
+      "learning_rate": 1.7958745163045986e-05,
+      "loss": 0.3218,
+      "step": 770
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.7961853742599487,
+      "learning_rate": 1.7841019623980217e-05,
+      "loss": 0.3709,
+      "step": 771
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6696933507919312,
+      "learning_rate": 1.77235973695485e-05,
+      "loss": 0.2508,
+      "step": 772
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6975236535072327,
+      "learning_rate": 1.760647950713566e-05,
+      "loss": 0.3638,
+      "step": 773
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6934524178504944,
+      "learning_rate": 1.74896671412558e-05,
+      "loss": 0.2799,
+      "step": 774
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.8749544024467468,
+      "learning_rate": 1.737316137354197e-05,
+      "loss": 0.4207,
+      "step": 775
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.737594485282898,
+      "learning_rate": 1.725696330273575e-05,
+      "loss": 0.3809,
+      "step": 776
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7070876955986023,
+      "learning_rate": 1.7141074024676912e-05,
+      "loss": 0.3071,
+      "step": 777
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6654912233352661,
+      "learning_rate": 1.7025494632293048e-05,
+      "loss": 0.2424,
+      "step": 778
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8109536170959473,
+      "learning_rate": 1.6910226215589303e-05,
+      "loss": 0.4148,
+      "step": 779
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7836554050445557,
+      "learning_rate": 1.6795269861638042e-05,
+      "loss": 0.3902,
+      "step": 780
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7711945176124573,
+      "learning_rate": 1.6680626654568686e-05,
+      "loss": 0.3066,
+      "step": 781
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.9572737812995911,
+      "learning_rate": 1.656629767555739e-05,
+      "loss": 0.4342,
+      "step": 782
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7705225348472595,
+      "learning_rate": 1.645228400281689e-05,
+      "loss": 0.3234,
+      "step": 783
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7431042790412903,
+      "learning_rate": 1.6338586711586356e-05,
+      "loss": 0.2927,
+      "step": 784
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.8527782559394836,
+      "learning_rate": 1.6225206874121218e-05,
+      "loss": 0.3685,
+      "step": 785
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.7141579985618591,
+      "learning_rate": 1.6112145559683055e-05,
+      "loss": 0.3218,
+      "step": 786
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.823011040687561,
+      "learning_rate": 1.5999403834529548e-05,
+      "loss": 0.4416,
+      "step": 787
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7237880229949951,
+      "learning_rate": 1.5886982761904377e-05,
+      "loss": 0.2902,
+      "step": 788
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8323907256126404,
+      "learning_rate": 1.5774883402027208e-05,
+      "loss": 0.3832,
+      "step": 789
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7947205901145935,
+      "learning_rate": 1.5663106812083743e-05,
+      "loss": 0.3388,
+      "step": 790
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7022159099578857,
+      "learning_rate": 1.555165404621567e-05,
+      "loss": 0.3068,
+      "step": 791
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7335956692695618,
+      "learning_rate": 1.5440526155510763e-05,
+      "loss": 0.3906,
+      "step": 792
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8507185578346252,
+      "learning_rate": 1.532972418799298e-05,
+      "loss": 0.4328,
+      "step": 793
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7950394153594971,
+      "learning_rate": 1.5219249188612556e-05,
+      "loss": 0.3196,
+      "step": 794
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.7802971005439758,
+      "learning_rate": 1.5109102199236152e-05,
+      "loss": 0.385,
+      "step": 795
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8072926998138428,
+      "learning_rate": 1.4999284258637053e-05,
+      "loss": 0.3626,
+      "step": 796
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7855871319770813,
+      "learning_rate": 1.488979640248534e-05,
+      "loss": 0.3309,
+      "step": 797
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7588586807250977,
+      "learning_rate": 1.4780639663338125e-05,
+      "loss": 0.2868,
+      "step": 798
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6957454681396484,
+      "learning_rate": 1.4671815070629868e-05,
+      "loss": 0.2715,
+      "step": 799
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6696025729179382,
+      "learning_rate": 1.4563323650662586e-05,
+      "loss": 0.2424,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.3238048255443573,
+      "eval_runtime": 372.3063,
+      "eval_samples_per_second": 0.916,
+      "eval_steps_per_second": 0.231,
+      "step": 800
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6485837697982788,
+      "learning_rate": 1.4455166426596223e-05,
+      "loss": 0.225,
+      "step": 801
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8903530836105347,
+      "learning_rate": 1.434734441843899e-05,
+      "loss": 0.535,
+      "step": 802
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8116011619567871,
+      "learning_rate": 1.4239858643037751e-05,
+      "loss": 0.333,
+      "step": 803
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6965086460113525,
+      "learning_rate": 1.4132710114068426e-05,
+      "loss": 0.3127,
+      "step": 804
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6642959117889404,
+      "learning_rate": 1.4025899842026442e-05,
+      "loss": 0.2486,
+      "step": 805
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7790483832359314,
+      "learning_rate": 1.3919428834217163e-05,
+      "loss": 0.2773,
+      "step": 806
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.7301218509674072,
+      "learning_rate": 1.3813298094746491e-05,
+      "loss": 0.2791,
+      "step": 807
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6421204805374146,
+      "learning_rate": 1.3707508624511262e-05,
+      "loss": 0.2376,
+      "step": 808
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7334969639778137,
+      "learning_rate": 1.36020614211899e-05,
+      "loss": 0.3691,
+      "step": 809
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6953814029693604,
+      "learning_rate": 1.349695747923298e-05,
+      "loss": 0.297,
+      "step": 810
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6431040167808533,
+      "learning_rate": 1.339219778985385e-05,
+      "loss": 0.3041,
+      "step": 811
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.83720862865448,
+      "learning_rate": 1.3287783341019278e-05,
+      "loss": 0.3345,
+      "step": 812
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7235584259033203,
+      "learning_rate": 1.3183715117440142e-05,
+      "loss": 0.3438,
+      "step": 813
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.868826687335968,
+      "learning_rate": 1.3079994100562159e-05,
+      "loss": 0.3742,
+      "step": 814
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7736153602600098,
+      "learning_rate": 1.2976621268556571e-05,
+      "loss": 0.3702,
+      "step": 815
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7487804293632507,
+      "learning_rate": 1.2873597596311027e-05,
+      "loss": 0.3076,
+      "step": 816
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.7739407420158386,
+      "learning_rate": 1.2770924055420258e-05,
+      "loss": 0.3598,
+      "step": 817
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.80709308385849,
+      "learning_rate": 1.2668601614177017e-05,
+      "loss": 0.4239,
+      "step": 818
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7017986178398132,
+      "learning_rate": 1.2566631237562893e-05,
+      "loss": 0.3084,
+      "step": 819
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7357339262962341,
+      "learning_rate": 1.246501388723923e-05,
+      "loss": 0.2905,
+      "step": 820
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7503935098648071,
+      "learning_rate": 1.2363750521538064e-05,
+      "loss": 0.2898,
+      "step": 821
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8020220398902893,
+      "learning_rate": 1.2262842095453065e-05,
+      "loss": 0.4112,
+      "step": 822
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.940615177154541,
+      "learning_rate": 1.2162289560630524e-05,
+      "loss": 0.3698,
+      "step": 823
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8232280015945435,
+      "learning_rate": 1.2062093865360458e-05,
+      "loss": 0.3915,
+      "step": 824
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6811102032661438,
+      "learning_rate": 1.1962255954567537e-05,
+      "loss": 0.2792,
+      "step": 825
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7059808373451233,
+      "learning_rate": 1.1862776769802276e-05,
+      "loss": 0.3096,
+      "step": 826
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.8150413632392883,
+      "learning_rate": 1.1763657249232107e-05,
+      "loss": 0.4002,
+      "step": 827
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6941980123519897,
+      "learning_rate": 1.1664898327632551e-05,
+      "loss": 0.2706,
+      "step": 828
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7593507766723633,
+      "learning_rate": 1.1566500936378388e-05,
+      "loss": 0.3814,
+      "step": 829
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.885452389717102,
+      "learning_rate": 1.146846600343488e-05,
+      "loss": 0.4113,
+      "step": 830
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.08356773853302,
+      "learning_rate": 1.1370794453349037e-05,
+      "loss": 0.4813,
+      "step": 831
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7590617537498474,
+      "learning_rate": 1.1273487207240847e-05,
+      "loss": 0.3094,
+      "step": 832
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6535730957984924,
+      "learning_rate": 1.1176545182794674e-05,
+      "loss": 0.2162,
+      "step": 833
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7632650136947632,
+      "learning_rate": 1.1079969294250514e-05,
+      "loss": 0.3224,
+      "step": 834
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7849392294883728,
+      "learning_rate": 1.0983760452395413e-05,
+      "loss": 0.3657,
+      "step": 835
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8800737261772156,
+      "learning_rate": 1.0887919564554894e-05,
+      "loss": 0.4074,
+      "step": 836
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8424676656723022,
+      "learning_rate": 1.079244753458437e-05,
+      "loss": 0.3857,
+      "step": 837
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.734151303768158,
+      "learning_rate": 1.0697345262860636e-05,
+      "loss": 0.3872,
+      "step": 838
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8216261863708496,
+      "learning_rate": 1.0602613646273374e-05,
+      "loss": 0.4122,
+      "step": 839
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8104004263877869,
+      "learning_rate": 1.0508253578216693e-05,
+      "loss": 0.4083,
+      "step": 840
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8514317274093628,
+      "learning_rate": 1.0414265948580692e-05,
+      "loss": 0.3823,
+      "step": 841
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7562483549118042,
+      "learning_rate": 1.032065164374313e-05,
+      "loss": 0.2868,
+      "step": 842
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6687902808189392,
+      "learning_rate": 1.0227411546560962e-05,
+      "loss": 0.2871,
+      "step": 843
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8124663829803467,
+      "learning_rate": 1.01345465363621e-05,
+      "loss": 0.3301,
+      "step": 844
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8382691144943237,
+      "learning_rate": 1.0042057488937067e-05,
+      "loss": 0.3337,
+      "step": 845
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.9069141149520874,
+      "learning_rate": 9.949945276530781e-06,
+      "loss": 0.4566,
+      "step": 846
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.681840181350708,
+      "learning_rate": 9.858210767834292e-06,
+      "loss": 0.282,
+      "step": 847
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8255618810653687,
+      "learning_rate": 9.766854827976617e-06,
+      "loss": 0.3169,
+      "step": 848
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8065633177757263,
+      "learning_rate": 9.675878318516546e-06,
+      "loss": 0.2903,
+      "step": 849
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.969791829586029,
+      "learning_rate": 9.58528209743459e-06,
+      "loss": 0.4463,
+      "step": 850
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8397563695907593,
+      "learning_rate": 9.495067019124792e-06,
+      "loss": 0.3544,
+      "step": 851
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.846594512462616,
+      "learning_rate": 9.405233934386726e-06,
+      "loss": 0.3263,
+      "step": 852
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.845501184463501,
+      "learning_rate": 9.315783690417479e-06,
+      "loss": 0.3022,
+      "step": 853
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.7699308395385742,
+      "learning_rate": 9.226717130803636e-06,
+      "loss": 0.287,
+      "step": 854
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.711404025554657,
+      "learning_rate": 9.138035095513336e-06,
+      "loss": 0.3685,
+      "step": 855
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6126477122306824,
+      "learning_rate": 9.049738420888349e-06,
+      "loss": 0.273,
+      "step": 856
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8038343787193298,
+      "learning_rate": 8.961827939636196e-06,
+      "loss": 0.3116,
+      "step": 857
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5792363286018372,
+      "learning_rate": 8.87430448082227e-06,
+      "loss": 0.1709,
+      "step": 858
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6694104671478271,
+      "learning_rate": 8.787168869862066e-06,
+      "loss": 0.2545,
+      "step": 859
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8506199717521667,
+      "learning_rate": 8.700421928513352e-06,
+      "loss": 0.4459,
+      "step": 860
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7557889223098755,
+      "learning_rate": 8.614064474868421e-06,
+      "loss": 0.3348,
+      "step": 861
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.884208619594574,
+      "learning_rate": 8.528097323346407e-06,
+      "loss": 0.3897,
+      "step": 862
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7232838273048401,
+      "learning_rate": 8.442521284685573e-06,
+      "loss": 0.2202,
+      "step": 863
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7185483574867249,
+      "learning_rate": 8.357337165935675e-06,
+      "loss": 0.3288,
+      "step": 864
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.9441823363304138,
+      "learning_rate": 8.27254577045039e-06,
+      "loss": 0.3954,
+      "step": 865
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7309289574623108,
+      "learning_rate": 8.188147897879667e-06,
+      "loss": 0.2677,
+      "step": 866
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8129590749740601,
+      "learning_rate": 8.104144344162229e-06,
+      "loss": 0.3022,
+      "step": 867
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8224502801895142,
+      "learning_rate": 8.02053590151805e-06,
+      "loss": 0.3059,
+      "step": 868
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8301034569740295,
+      "learning_rate": 7.937323358440935e-06,
+      "loss": 0.3438,
+      "step": 869
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.6451743245124817,
+      "learning_rate": 7.854507499691005e-06,
+      "loss": 0.2137,
+      "step": 870
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8484547734260559,
+      "learning_rate": 7.772089106287344e-06,
+      "loss": 0.3504,
+      "step": 871
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8294460773468018,
+      "learning_rate": 7.690068955500624e-06,
+      "loss": 0.4083,
+      "step": 872
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7790281772613525,
+      "learning_rate": 7.608447820845771e-06,
+      "loss": 0.3122,
+      "step": 873
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7521539330482483,
+      "learning_rate": 7.527226472074678e-06,
+      "loss": 0.3169,
+      "step": 874
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7376779913902283,
+      "learning_rate": 7.446405675168938e-06,
+      "loss": 0.3659,
+      "step": 875
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.6949995756149292,
+      "learning_rate": 7.365986192332625e-06,
+      "loss": 0.2257,
+      "step": 876
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7358771562576294,
+      "learning_rate": 7.285968781985092e-06,
+      "loss": 0.3131,
+      "step": 877
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7232546210289001,
+      "learning_rate": 7.206354198753862e-06,
+      "loss": 0.2336,
+      "step": 878
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8304458260536194,
+      "learning_rate": 7.127143193467445e-06,
+      "loss": 0.3829,
+      "step": 879
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.9551928043365479,
+      "learning_rate": 7.048336513148307e-06,
+      "loss": 0.4036,
+      "step": 880
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.98590487241745,
+      "learning_rate": 6.969934901005809e-06,
+      "loss": 0.5269,
+      "step": 881
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8517870903015137,
+      "learning_rate": 6.8919390964291895e-06,
+      "loss": 0.3954,
+      "step": 882
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7202636003494263,
+      "learning_rate": 6.814349834980621e-06,
+      "loss": 0.3277,
+      "step": 883
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8719897866249084,
+      "learning_rate": 6.7371678483882264e-06,
+      "loss": 0.4527,
+      "step": 884
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6927401423454285,
+      "learning_rate": 6.660393864539222e-06,
+      "loss": 0.2947,
+      "step": 885
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6914498805999756,
+      "learning_rate": 6.58402860747302e-06,
+      "loss": 0.2437,
+      "step": 886
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8315032720565796,
+      "learning_rate": 6.508072797374454e-06,
+      "loss": 0.3975,
+      "step": 887
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7579956650733948,
+      "learning_rate": 6.4325271505669025e-06,
+      "loss": 0.3591,
+      "step": 888
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8064132332801819,
+      "learning_rate": 6.3573923795056e-06,
+      "loss": 0.3456,
+      "step": 889
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9116144180297852,
+      "learning_rate": 6.282669192770896e-06,
+      "loss": 0.4022,
+      "step": 890
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8875893950462341,
+      "learning_rate": 6.208358295061572e-06,
+      "loss": 0.3727,
+      "step": 891
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6741305589675903,
+      "learning_rate": 6.134460387188206e-06,
+      "loss": 0.2776,
+      "step": 892
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6826175451278687,
+      "learning_rate": 6.060976166066546e-06,
+      "loss": 0.2993,
+      "step": 893
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7811982035636902,
+      "learning_rate": 5.9879063247109405e-06,
+      "loss": 0.3785,
+      "step": 894
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8341353535652161,
+      "learning_rate": 5.915251552227829e-06,
+      "loss": 0.3576,
+      "step": 895
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.2248538732528687,
+      "learning_rate": 5.8430125338092115e-06,
+      "loss": 0.4661,
+      "step": 896
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.8265413045883179,
+      "learning_rate": 5.771189950726191e-06,
+      "loss": 0.2916,
+      "step": 897
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9220693111419678,
+      "learning_rate": 5.699784480322568e-06,
+      "loss": 0.4308,
+      "step": 898
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7422018647193909,
+      "learning_rate": 5.628796796008434e-06,
+      "loss": 0.3689,
+      "step": 899
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7564661502838135,
+      "learning_rate": 5.558227567253832e-06,
+      "loss": 0.328,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.3223819434642792,
+      "eval_runtime": 372.4052,
+      "eval_samples_per_second": 0.916,
+      "eval_steps_per_second": 0.231,
+      "step": 900
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7492172718048096,
+      "learning_rate": 5.488077459582425e-06,
+      "loss": 0.3625,
+      "step": 901
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6984173059463501,
+      "learning_rate": 5.418347134565249e-06,
+      "loss": 0.3503,
+      "step": 902
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7128195762634277,
+      "learning_rate": 5.349037249814442e-06,
+      "loss": 0.3173,
+      "step": 903
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7856778502464294,
+      "learning_rate": 5.28014845897708e-06,
+      "loss": 0.3607,
+      "step": 904
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8123862147331238,
+      "learning_rate": 5.211681411728969e-06,
+      "loss": 0.3764,
+      "step": 905
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8320104479789734,
+      "learning_rate": 5.143636753768549e-06,
+      "loss": 0.3068,
+      "step": 906
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7244311571121216,
+      "learning_rate": 5.076015126810784e-06,
+      "loss": 0.3074,
+      "step": 907
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7522035837173462,
+      "learning_rate": 5.008817168581137e-06,
+      "loss": 0.3168,
+      "step": 908
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8384779095649719,
+      "learning_rate": 4.94204351280953e-06,
+      "loss": 0.3587,
+      "step": 909
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8368098735809326,
+      "learning_rate": 4.8756947892243725e-06,
+      "loss": 0.3898,
+      "step": 910
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.7449876070022583,
+      "learning_rate": 4.809771623546627e-06,
+      "loss": 0.2622,
+      "step": 911
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.657548189163208,
+      "learning_rate": 4.744274637483936e-06,
+      "loss": 0.2671,
+      "step": 912
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7042245268821716,
+      "learning_rate": 4.6792044487247e-06,
+      "loss": 0.332,
+      "step": 913
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7108582854270935,
+      "learning_rate": 4.614561670932288e-06,
+      "loss": 0.3143,
+      "step": 914
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.7174756526947021,
+      "learning_rate": 4.550346913739256e-06,
+      "loss": 0.2832,
+      "step": 915
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6656290292739868,
+      "learning_rate": 4.486560782741578e-06,
+      "loss": 0.2533,
+      "step": 916
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.593611478805542,
+      "learning_rate": 4.423203879492943e-06,
+      "loss": 0.2199,
+      "step": 917
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8348184823989868,
+      "learning_rate": 4.360276801499086e-06,
+      "loss": 0.4036,
+      "step": 918
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8101851940155029,
+      "learning_rate": 4.29778014221216e-06,
+      "loss": 0.3314,
+      "step": 919
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8485448360443115,
+      "learning_rate": 4.2357144910251e-06,
+      "loss": 0.3651,
+      "step": 920
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.9444641470909119,
+      "learning_rate": 4.174080433266136e-06,
+      "loss": 0.3868,
+      "step": 921
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.755634069442749,
+      "learning_rate": 4.112878550193194e-06,
+      "loss": 0.3814,
+      "step": 922
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.6165770292282104,
+      "learning_rate": 4.05210941898847e-06,
+      "loss": 0.2395,
+      "step": 923
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7212772965431213,
+      "learning_rate": 3.991773612752952e-06,
+      "loss": 0.2933,
+      "step": 924
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7796584367752075,
+      "learning_rate": 3.9318717005010495e-06,
+      "loss": 0.3775,
+      "step": 925
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7791245579719543,
+      "learning_rate": 3.872404247155193e-06,
+      "loss": 0.3643,
+      "step": 926
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8050986528396606,
+      "learning_rate": 3.8133718135405285e-06,
+      "loss": 0.3981,
+      "step": 927
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.7599121332168579,
+      "learning_rate": 3.754774956379614e-06,
+      "loss": 0.3321,
+      "step": 928
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8776902556419373,
+      "learning_rate": 3.696614228287187e-06,
+      "loss": 0.3786,
+      "step": 929
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.868802011013031,
+      "learning_rate": 3.6388901777649477e-06,
+      "loss": 0.3645,
+      "step": 930
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8056930303573608,
+      "learning_rate": 3.581603349196372e-06,
+      "loss": 0.3867,
+      "step": 931
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8505524396896362,
+      "learning_rate": 3.5247542828415747e-06,
+      "loss": 0.4482,
+      "step": 932
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.652051568031311,
+      "learning_rate": 3.468343514832251e-06,
+      "loss": 0.2632,
+      "step": 933
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.6999272704124451,
+      "learning_rate": 3.4123715771665786e-06,
+      "loss": 0.2289,
+      "step": 934
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.795468270778656,
+      "learning_rate": 3.356838997704226e-06,
+      "loss": 0.3945,
+      "step": 935
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7194265127182007,
+      "learning_rate": 3.3017463001613626e-06,
+      "loss": 0.312,
+      "step": 936
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6935940980911255,
+      "learning_rate": 3.2470940041057106e-06,
+      "loss": 0.2665,
+      "step": 937
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6907996535301208,
+      "learning_rate": 3.1928826249516987e-06,
+      "loss": 0.2501,
+      "step": 938
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.7125189304351807,
+      "learning_rate": 3.139112673955513e-06,
+      "loss": 0.3341,
+      "step": 939
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.7351633310317993,
+      "learning_rate": 3.08578465821035e-06,
+      "loss": 0.3409,
+      "step": 940
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6781859397888184,
+      "learning_rate": 3.0328990806415934e-06,
+      "loss": 0.3027,
+      "step": 941
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.9054991006851196,
+      "learning_rate": 2.9804564400020994e-06,
+      "loss": 0.3926,
+      "step": 942
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8243528008460999,
+      "learning_rate": 2.9284572308674596e-06,
+      "loss": 0.403,
+      "step": 943
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8222914934158325,
+      "learning_rate": 2.8769019436313715e-06,
+      "loss": 0.3623,
+      "step": 944
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7432158589363098,
+      "learning_rate": 2.825791064500993e-06,
+      "loss": 0.3255,
+      "step": 945
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7823918461799622,
+      "learning_rate": 2.7751250754923573e-06,
+      "loss": 0.3406,
+      "step": 946
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7455360889434814,
+      "learning_rate": 2.7249044544258363e-06,
+      "loss": 0.3735,
+      "step": 947
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8658504486083984,
+      "learning_rate": 2.675129674921639e-06,
+      "loss": 0.5029,
+      "step": 948
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8966514468193054,
+      "learning_rate": 2.625801206395312e-06,
+      "loss": 0.4415,
+      "step": 949
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7794380784034729,
+      "learning_rate": 2.576919514053355e-06,
+      "loss": 0.3122,
+      "step": 950
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7666850090026855,
+      "learning_rate": 2.528485058888813e-06,
+      "loss": 0.3979,
+      "step": 951
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8177826404571533,
+      "learning_rate": 2.48049829767692e-06,
+      "loss": 0.3946,
+      "step": 952
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9126067757606506,
+      "learning_rate": 2.4329596829708144e-06,
+      "loss": 0.4382,
+      "step": 953
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6944770216941833,
+      "learning_rate": 2.385869663097251e-06,
+      "loss": 0.3095,
+      "step": 954
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9347777962684631,
+      "learning_rate": 2.3392286821523724e-06,
+      "loss": 0.414,
+      "step": 955
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6874916553497314,
+      "learning_rate": 2.2930371799975594e-06,
+      "loss": 0.254,
+      "step": 956
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6940197944641113,
+      "learning_rate": 2.2472955922552163e-06,
+      "loss": 0.288,
+      "step": 957
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8366566896438599,
+      "learning_rate": 2.202004350304715e-06,
+      "loss": 0.3972,
+      "step": 958
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8117697238922119,
+      "learning_rate": 2.157163881278312e-06,
+      "loss": 0.3753,
+      "step": 959
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7425122261047363,
+      "learning_rate": 2.112774608057111e-06,
+      "loss": 0.3204,
+      "step": 960
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.120468258857727,
+      "learning_rate": 2.068836949267089e-06,
+      "loss": 0.382,
+      "step": 961
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7509793043136597,
+      "learning_rate": 2.0253513192751373e-06,
+      "loss": 0.3209,
+      "step": 962
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7593684196472168,
+      "learning_rate": 1.982318128185151e-06,
+      "loss": 0.3542,
+      "step": 963
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.8635910153388977,
+      "learning_rate": 1.9397377818341944e-06,
+      "loss": 0.3665,
+      "step": 964
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7751469612121582,
+      "learning_rate": 1.8976106817886196e-06,
+      "loss": 0.4008,
+      "step": 965
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8273268938064575,
+      "learning_rate": 1.855937225340315e-06,
+      "loss": 0.3425,
+      "step": 966
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7622714042663574,
+      "learning_rate": 1.8147178055029579e-06,
+      "loss": 0.3121,
+      "step": 967
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6592556238174438,
+      "learning_rate": 1.7739528110083004e-06,
+      "loss": 0.2856,
+      "step": 968
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.775131106376648,
+      "learning_rate": 1.7336426263024897e-06,
+      "loss": 0.3438,
+      "step": 969
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6192392706871033,
+      "learning_rate": 1.6937876315424705e-06,
+      "loss": 0.2112,
+      "step": 970
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6981117129325867,
+      "learning_rate": 1.6543882025923886e-06,
+      "loss": 0.3059,
+      "step": 971
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7565510869026184,
+      "learning_rate": 1.6154447110200255e-06,
+      "loss": 0.3524,
+      "step": 972
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8459018468856812,
+      "learning_rate": 1.5769575240933422e-06,
+      "loss": 0.2848,
+      "step": 973
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9841071367263794,
+      "learning_rate": 1.5389270047769578e-06,
+      "loss": 0.5358,
+      "step": 974
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8021631240844727,
+      "learning_rate": 1.5013535117287647e-06,
+      "loss": 0.3447,
+      "step": 975
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.710159957408905,
+      "learning_rate": 1.4642373992965364e-06,
+      "loss": 0.2492,
+      "step": 976
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7437825202941895,
+      "learning_rate": 1.4275790175145908e-06,
+      "loss": 0.2798,
+      "step": 977
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7793743014335632,
+      "learning_rate": 1.3913787121004719e-06,
+      "loss": 0.3217,
+      "step": 978
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8428955674171448,
+      "learning_rate": 1.3556368244517114e-06,
+      "loss": 0.365,
+      "step": 979
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6851413249969482,
+      "learning_rate": 1.3203536916425841e-06,
+      "loss": 0.3022,
+      "step": 980
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7840973138809204,
+      "learning_rate": 1.2855296464209688e-06,
+      "loss": 0.3559,
+      "step": 981
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8621905446052551,
+      "learning_rate": 1.2511650172051637e-06,
+      "loss": 0.474,
+      "step": 982
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7440460324287415,
+      "learning_rate": 1.217260128080816e-06,
+      "loss": 0.3349,
+      "step": 983
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.6954957246780396,
+      "learning_rate": 1.1838152987978578e-06,
+      "loss": 0.3184,
+      "step": 984
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7392785549163818,
+      "learning_rate": 1.1508308447674975e-06,
+      "loss": 0.3454,
+      "step": 985
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.7514981031417847,
+      "learning_rate": 1.1183070770592441e-06,
+      "loss": 0.3516,
+      "step": 986
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7906771302223206,
+      "learning_rate": 1.0862443023979651e-06,
+      "loss": 0.3041,
+      "step": 987
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7142906188964844,
+      "learning_rate": 1.0546428231609895e-06,
+      "loss": 0.2924,
+      "step": 988
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.9594712853431702,
+      "learning_rate": 1.0235029373752758e-06,
+      "loss": 0.4788,
+      "step": 989
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7356728315353394,
+      "learning_rate": 9.928249387145982e-07,
+      "loss": 0.3055,
+      "step": 990
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7300496697425842,
+      "learning_rate": 9.6260911649676e-07,
+      "loss": 0.274,
+      "step": 991
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6496745347976685,
+      "learning_rate": 9.32855755680867e-07,
+      "loss": 0.2776,
+      "step": 992
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.7497766017913818,
+      "learning_rate": 9.035651368646648e-07,
+      "loss": 0.3507,
+      "step": 993
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.9220755100250244,
+      "learning_rate": 8.747375362818666e-07,
+      "loss": 0.4941,
+      "step": 994
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8557547330856323,
+      "learning_rate": 8.463732257995571e-07,
+      "loss": 0.3848,
+      "step": 995
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.6327918171882629,
+      "learning_rate": 8.184724729156379e-07,
+      "loss": 0.23,
+      "step": 996
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8224833011627197,
+      "learning_rate": 7.910355407562742e-07,
+      "loss": 0.3483,
+      "step": 997
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7008509635925293,
+      "learning_rate": 7.640626880734581e-07,
+      "loss": 0.2764,
+      "step": 998
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6525281071662903,
+      "learning_rate": 7.375541692425326e-07,
+      "loss": 0.2355,
+      "step": 999
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8102105855941772,
+      "learning_rate": 7.115102342598101e-07,
+      "loss": 0.3847,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.3205099105834961,
+      "eval_runtime": 359.4096,
+      "eval_samples_per_second": 0.949,
+      "eval_steps_per_second": 0.239,
+      "step": 1000
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7621821761131287,
+      "learning_rate": 6.859311287402081e-07,
+      "loss": 0.2805,
+      "step": 1001
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7008903622627258,
+      "learning_rate": 6.608170939149283e-07,
+      "loss": 0.2357,
+      "step": 1002
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8182081580162048,
+      "learning_rate": 6.361683666291973e-07,
+      "loss": 0.3705,
+      "step": 1003
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.77609783411026,
+      "learning_rate": 6.119851793400189e-07,
+      "loss": 0.344,
+      "step": 1004
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7903171181678772,
+      "learning_rate": 5.882677601139919e-07,
+      "loss": 0.3147,
+      "step": 1005
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7790983319282532,
+      "learning_rate": 5.650163326251345e-07,
+      "loss": 0.3009,
+      "step": 1006
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9050146341323853,
+      "learning_rate": 5.422311161528193e-07,
+      "loss": 0.4384,
+      "step": 1007
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.9335758090019226,
+      "learning_rate": 5.199123255796634e-07,
+      "loss": 0.3528,
+      "step": 1008
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6988196969032288,
+      "learning_rate": 4.980601713895305e-07,
+      "loss": 0.2112,
+      "step": 1009
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.746230959892273,
+      "learning_rate": 4.7667485966552685e-07,
+      "loss": 0.2757,
+      "step": 1010
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8325840830802917,
+      "learning_rate": 4.55756592088058e-07,
+      "loss": 0.3587,
+      "step": 1011
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.7857056260108948,
+      "learning_rate": 4.353055659329419e-07,
+      "loss": 0.4106,
+      "step": 1012
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.0377814769744873,
+      "learning_rate": 4.153219740695435e-07,
+      "loss": 0.4745,
+      "step": 1013
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.9208704233169556,
+      "learning_rate": 3.958060049589485e-07,
+      "loss": 0.4507,
+      "step": 1014
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6509748697280884,
+      "learning_rate": 3.767578426521923e-07,
+      "loss": 0.2706,
+      "step": 1015
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6952648758888245,
+      "learning_rate": 3.5817766678850615e-07,
+      "loss": 0.2881,
+      "step": 1016
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.783473789691925,
+      "learning_rate": 3.4006565259366295e-07,
+      "loss": 0.284,
+      "step": 1017
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8076602220535278,
+      "learning_rate": 3.224219708782894e-07,
+      "loss": 0.3364,
+      "step": 1018
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.0560907125473022,
+      "learning_rate": 3.052467880362675e-07,
+      "loss": 0.4805,
+      "step": 1019
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.768387496471405,
+      "learning_rate": 2.88540266043158e-07,
+      "loss": 0.3608,
+      "step": 1020
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8374309539794922,
+      "learning_rate": 2.723025624546849e-07,
+      "loss": 0.3666,
+      "step": 1021
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.774454653263092,
+      "learning_rate": 2.5653383040524227e-07,
+      "loss": 0.3597,
+      "step": 1022
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8575447797775269,
+      "learning_rate": 2.412342186064565e-07,
+      "loss": 0.2638,
+      "step": 1023
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8327907919883728,
+      "learning_rate": 2.2640387134577058e-07,
+      "loss": 0.4489,
+      "step": 1024
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7777142524719238,
+      "learning_rate": 2.1204292848509555e-07,
+      "loss": 0.3197,
+      "step": 1025
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.7611891031265259,
+      "learning_rate": 1.98151525459489e-07,
+      "loss": 0.2947,
+      "step": 1026
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.6954681873321533,
+      "learning_rate": 1.8472979327587292e-07,
+      "loss": 0.2649,
+      "step": 1027
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9290033578872681,
+      "learning_rate": 1.717778585118013e-07,
+      "loss": 0.4053,
+      "step": 1028
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7652479410171509,
+      "learning_rate": 1.5929584331427216e-07,
+      "loss": 0.35,
+      "step": 1029
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7672208547592163,
+      "learning_rate": 1.4728386539856754e-07,
+      "loss": 0.3857,
+      "step": 1030
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8665684461593628,
+      "learning_rate": 1.3574203804713747e-07,
+      "loss": 0.4668,
+      "step": 1031
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7497051954269409,
+      "learning_rate": 1.246704701085566e-07,
+      "loss": 0.348,
+      "step": 1032
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.745912492275238,
+      "learning_rate": 1.1406926599646373e-07,
+      "loss": 0.3008,
+      "step": 1033
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7423427104949951,
+      "learning_rate": 1.0393852568860719e-07,
+      "loss": 0.2818,
+      "step": 1034
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6192858219146729,
+      "learning_rate": 9.427834472588992e-08,
+      "loss": 0.1777,
+      "step": 1035
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8588548898696899,
+      "learning_rate": 8.508881421145365e-08,
+      "loss": 0.3109,
+      "step": 1036
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7353888750076294,
+      "learning_rate": 7.637002080985168e-08,
+      "loss": 0.2928,
+      "step": 1037
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8530696034431458,
+      "learning_rate": 6.812204674619959e-08,
+      "loss": 0.3488,
+      "step": 1038
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.6897252202033997,
+      "learning_rate": 6.034496980542037e-08,
+      "loss": 0.3031,
+      "step": 1039
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.765937089920044,
+      "learning_rate": 5.303886333151153e-08,
+      "loss": 0.3346,
+      "step": 1040
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7663593888282776,
+      "learning_rate": 4.620379622682358e-08,
+      "loss": 0.343,
+      "step": 1041
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7448627352714539,
+      "learning_rate": 3.9839832951465986e-08,
+      "loss": 0.297,
+      "step": 1042
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.861727237701416,
+      "learning_rate": 3.394703352263551e-08,
+      "loss": 0.3186,
+      "step": 1043
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8436367511749268,
+      "learning_rate": 2.8525453514099965e-08,
+      "loss": 0.3996,
+      "step": 1044
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.6379160284996033,
+      "learning_rate": 2.3575144055643094e-08,
+      "loss": 0.2129,
+      "step": 1045
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7607189416885376,
+      "learning_rate": 1.9096151832609375e-08,
+      "loss": 0.3138,
+      "step": 1046
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8000789880752563,
+      "learning_rate": 1.5088519085437734e-08,
+      "loss": 0.3627,
+      "step": 1047
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7188413143157959,
+      "learning_rate": 1.1552283609272962e-08,
+      "loss": 0.2724,
+      "step": 1048
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.7416312098503113,
+      "learning_rate": 8.487478753615997e-09,
+      "loss": 0.3899,
+      "step": 1049
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8885815739631653,
+      "learning_rate": 5.894133422001957e-09,
+      "loss": 0.4372,
+      "step": 1050
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6155141592025757,
+      "learning_rate": 3.772272071722593e-09,
+      "loss": 0.2539,
+      "step": 1051
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7091107368469238,
+      "learning_rate": 2.1219147136264382e-09,
+      "loss": 0.2898,
+      "step": 1052
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7850836515426636,
+      "learning_rate": 9.43076911874563e-10,
+      "loss": 0.4009,
+      "step": 1053
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8586174249649048,
+      "learning_rate": 2.3576978384065584e-10,
+      "loss": 0.3486,
+      "step": 1054
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7261866927146912,
+      "learning_rate": 0.0,
+      "loss": 0.3182,
+      "step": 1055
+    },
+    {
+      "epoch": 1.0,
+      "step": 1055,
+      "total_flos": 1.3702264916017152e+16,
+      "train_loss": 0.35872868220388043,
+      "train_runtime": 46682.626,
+      "train_samples_per_second": 0.723,
+      "train_steps_per_second": 0.023
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1055,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.3702264916017152e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}