vision_on

README.md

@@ -46,7 +46,7 @@ The following hyperparameters were used during training:
 - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 1.0
+- num_epochs: 15.0
 - mixed_precision_training: Native AMP
 
 ### Training results

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.0,
-    "total_flos": 2315448779538432.0,
-    "train_loss": 1.0368848323822022,
-    "train_runtime": 182.4958,
-    "train_samples_per_second": 1.096,
-    "train_steps_per_second": 0.548
+    "epoch": 15.0,
+    "total_flos": 8.815859931388314e+16,
+    "train_loss": 0.5605581335703532,
+    "train_runtime": 6084.0454,
+    "train_samples_per_second": 1.233,
+    "train_steps_per_second": 0.616
 }

checkpoint-3750/added_tokens.json

@@ -0,0 +1,16 @@
+{
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

checkpoint-3750/chat_template.json

@@ -0,0 +1,3 @@
+{
+  "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}"
+}

checkpoint-3750/config.json

@@ -0,0 +1,49 @@
+{
+  "_name_or_path": "Qwen/Qwen2-VL-2B-Instruct",
+  "architectures": [
+    "Qwen2VLForConditionalGeneration"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "image_token_id": 151655,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen2_vl",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": {
+    "mrope_section": [
+      16,
+      24,
+      24
+    ],
+    "rope_type": "default",
+    "type": "default"
+  },
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "tie_word_embeddings": true,
+  "torch_dtype": "float16",
+  "transformers_version": "4.49.0",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "video_token_id": 151656,
+  "vision_config": {
+    "hidden_size": 1536,
+    "in_chans": 3,
+    "model_type": "qwen2_vl",
+    "spatial_patch_size": 14,
+    "torch_dtype": "float16"
+  },
+  "vision_end_token_id": 151653,
+  "vision_start_token_id": 151652,
+  "vision_token_id": 151654,
+  "vocab_size": 151936
+}

checkpoint-3750/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.01,
+  "top_k": 1,
+  "top_p": 0.001,
+  "transformers_version": "4.49.0"
+}

checkpoint-3750/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:26a8460fd5a0de85bee5542e09267a1fadbf54411155a8aa88b01bdbc7daf62d
+size 4486225984

checkpoint-3750/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd33fc6c9022306092336d740089b98dd2340136b9419d8a36f7c405571c29fc
+size 272709388

checkpoint-3750/preprocessor_config.json

@@ -0,0 +1,29 @@
+{
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_processor_type": "Qwen2VLImageProcessor",
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "max_pixels": 12845056,
+  "merge_size": 2,
+  "min_pixels": 3136,
+  "patch_size": 14,
+  "processor_class": "Qwen2VLProcessor",
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "longest_edge": 12845056,
+    "shortest_edge": 3136
+  },
+  "temporal_patch_size": 2
+}

checkpoint-3750/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b72d3ff0d9cef769a20852fa3473da68a2ad59950ff6438edf98c95328ba5c1
+size 14512

checkpoint-3750/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bff6047902e34b76f0af39d44c10111f1acce1cfd48509bfb6b92e02a5c71d0f
+size 14512

checkpoint-3750/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db605402fddf58d0e4e0d0b239226846ecd7aede147e6a2db0c56c7ac94fea9a
+size 988

checkpoint-3750/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d4f458d78e801c33a68cdfa5e650b8e127a2331e36e35e26e658e885beeeb0c
+size 1064

checkpoint-3750/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-3750/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:091aa7594dc2fcfbfa06b9e3c22a5f0562ac14f30375c13af7309407a0e67b8a
+size 11420371

checkpoint-3750/tokenizer_config.json

@@ -0,0 +1,145 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% for message in messages %}{% if loop.first and message['role'] != 'system' %}<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n{% endif %}<|im_start|>{{ message['role'] }}\n{% if message['content'] is string %}{{ message['content'] }}<|im_end|>\n{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% if add_vision_id %}Picture {{ image_count.value }}: {% endif %}<|vision_start|><|image_pad|><|vision_end|>{% elif content['type'] == 'video' or 'video' in content %}{% set video_count.value = video_count.value + 1 %}{% if add_vision_id %}Video {{ video_count.value }}: {% endif %}<|vision_start|><|video_pad|><|vision_end|>{% elif 'text' in content %}{{ content['text'] }}{% endif %}{% endfor %}<|im_end|>\n{% endif %}{% endfor %}{% if add_generation_prompt %}<|im_start|>assistant\n{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 2048,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "processor_class": "Qwen2VLProcessor",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

checkpoint-3750/trainer_state.json

@@ -0,0 +1,2658 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 15.0,
+  "eval_steps": 500,
+  "global_step": 3750,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.0283896923065186,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 1.5494,
+      "step": 10
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.980957508087158,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 1.5781,
+      "step": 20
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.3325070142745972,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.271,
+      "step": 30
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1161530017852783,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 1.1263,
+      "step": 40
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9847109317779541,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.4759,
+      "step": 50
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.3176578283309937,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.1906,
+      "step": 60
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.9614243507385254,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 1.3136,
+      "step": 70
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.0635404586791992,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 1.159,
+      "step": 80
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4873822927474976,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.0645,
+      "step": 90
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.666663646697998,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.3174,
+      "step": 100
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9259383678436279,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 1.0041,
+      "step": 110
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.422500729560852,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.1208,
+      "step": 120
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.513822317123413,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 0.9806,
+      "step": 130
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.26381254196167,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 0.9904,
+      "step": 140
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.899075984954834,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.9365,
+      "step": 150
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7061178088188171,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 0.8843,
+      "step": 160
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.0236766338348389,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 0.9131,
+      "step": 170
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.3964245319366455,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.9142,
+      "step": 180
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7908815145492554,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 0.8973,
+      "step": 190
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.9264830946922302,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.8407,
+      "step": 200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5779850482940674,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.8228,
+      "step": 210
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6980341672897339,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 0.799,
+      "step": 220
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9194531440734863,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 0.7759,
+      "step": 230
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.0350003242492676,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.7466,
+      "step": 240
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 3.474932909011841,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7317,
+      "step": 250
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.7944034337997437,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 0.7046,
+      "step": 260
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.945058584213257,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.6625,
+      "step": 270
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.820989966392517,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 0.7307,
+      "step": 280
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.9544613361358643,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 0.7996,
+      "step": 290
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9499707221984863,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.706,
+      "step": 300
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 5.53220796585083,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 0.6941,
+      "step": 310
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.7142622470855713,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 0.746,
+      "step": 320
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 4.010003089904785,
+      "learning_rate": 1.76e-05,
+      "loss": 0.7262,
+      "step": 330
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 2.3094098567962646,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 0.6552,
+      "step": 340
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 5.371938228607178,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.677,
+      "step": 350
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.662387728691101,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.6551,
+      "step": 360
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.2856159210205078,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 0.7076,
+      "step": 370
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2962098121643066,
+      "learning_rate": 1.999989169177959e-05,
+      "loss": 0.6818,
+      "step": 380
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.3778997659683228,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 0.6823,
+      "step": 390
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.4772292375564575,
+      "learning_rate": 1.999729241179462e-05,
+      "loss": 0.7704,
+      "step": 400
+    },
+    {
+      "epoch": 1.6400000000000001,
+      "grad_norm": 1.13938307762146,
+      "learning_rate": 1.999469335702714e-05,
+      "loss": 0.6668,
+      "step": 410
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 2.4890644550323486,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.6448,
+      "step": 420
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.2066543102264404,
+      "learning_rate": 1.998689754391257e-05,
+      "loss": 0.7159,
+      "step": 430
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.4606579542160034,
+      "learning_rate": 1.998170146104234e-05,
+      "loss": 0.6443,
+      "step": 440
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 5.692836284637451,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 0.7253,
+      "step": 450
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.7549424171447754,
+      "learning_rate": 1.9968715193738738e-05,
+      "loss": 0.6349,
+      "step": 460
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.5602545738220215,
+      "learning_rate": 1.9960926134514875e-05,
+      "loss": 0.6793,
+      "step": 470
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.393797755241394,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 0.6686,
+      "step": 480
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.6151896119117737,
+      "learning_rate": 1.9942759539322845e-05,
+      "loss": 0.6584,
+      "step": 490
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.0221006870269775,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.6771,
+      "step": 500
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.0078063011169434,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.6664,
+      "step": 510
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.788282871246338,
+      "learning_rate": 1.990905082010344e-05,
+      "loss": 0.6243,
+      "step": 520
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.064715623855591,
+      "learning_rate": 1.9896096046383456e-05,
+      "loss": 0.6253,
+      "step": 530
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.5293374061584473,
+      "learning_rate": 1.988228381446553e-05,
+      "loss": 0.6362,
+      "step": 540
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.461493730545044,
+      "learning_rate": 1.9867615321125796e-05,
+      "loss": 0.6517,
+      "step": 550
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.1433868408203125,
+      "learning_rate": 1.985209183733209e-05,
+      "loss": 0.6849,
+      "step": 560
+    },
+    {
+      "epoch": 2.2800000000000002,
+      "grad_norm": 1.6532901525497437,
+      "learning_rate": 1.983571470813386e-05,
+      "loss": 0.6298,
+      "step": 570
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 3.705383539199829,
+      "learning_rate": 1.9818485352545595e-05,
+      "loss": 0.6588,
+      "step": 580
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.4615492820739746,
+      "learning_rate": 1.980040526342388e-05,
+      "loss": 0.6154,
+      "step": 590
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.8189066052436829,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.6393,
+      "step": 600
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.1122651100158691,
+      "learning_rate": 1.9761699224434476e-05,
+      "loss": 0.6245,
+      "step": 610
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.4684017896652222,
+      "learning_rate": 1.9741076628294387e-05,
+      "loss": 0.6592,
+      "step": 620
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.9914065599441528,
+      "learning_rate": 1.9719610005785466e-05,
+      "loss": 0.6262,
+      "step": 630
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.7366482019424438,
+      "learning_rate": 1.969730121690698e-05,
+      "loss": 0.672,
+      "step": 640
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 3.544377326965332,
+      "learning_rate": 1.967415219462864e-05,
+      "loss": 0.6057,
+      "step": 650
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.9553754329681396,
+      "learning_rate": 1.9650164944723116e-05,
+      "loss": 0.6142,
+      "step": 660
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 2.1661672592163086,
+      "learning_rate": 1.9625341545592226e-05,
+      "loss": 0.6238,
+      "step": 670
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "grad_norm": 3.7167468070983887,
+      "learning_rate": 1.9599684148086876e-05,
+      "loss": 0.7166,
+      "step": 680
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.688824415206909,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.6769,
+      "step": 690
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 4.10930061340332,
+      "learning_rate": 1.954587632247732e-05,
+      "loss": 0.6199,
+      "step": 700
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.5201390981674194,
+      "learning_rate": 1.951773055661174e-05,
+      "loss": 0.6242,
+      "step": 710
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 3.6892731189727783,
+      "learning_rate": 1.9488760116444966e-05,
+      "loss": 0.6245,
+      "step": 720
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.8859150409698486,
+      "learning_rate": 1.9458967512152872e-05,
+      "loss": 0.628,
+      "step": 730
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.4320142269134521,
+      "learning_rate": 1.9428355325148632e-05,
+      "loss": 0.5806,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.4816261529922485,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.5818,
+      "step": 750
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 2.1367580890655518,
+      "learning_rate": 1.9364682883494892e-05,
+      "loss": 0.6387,
+      "step": 760
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 2.2321407794952393,
+      "learning_rate": 1.9331628145814587e-05,
+      "loss": 0.6207,
+      "step": 770
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 3.410268783569336,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.5868,
+      "step": 780
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 3.323219060897827,
+      "learning_rate": 1.926309595682066e-05,
+      "loss": 0.5444,
+      "step": 790
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 2.397799015045166,
+      "learning_rate": 1.9227624443554425e-05,
+      "loss": 0.5891,
+      "step": 800
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 9.090506553649902,
+      "learning_rate": 1.9191353392552346e-05,
+      "loss": 0.5453,
+      "step": 810
+    },
+    {
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.3556101322174072,
+      "learning_rate": 1.9154285946559792e-05,
+      "loss": 0.6406,
+      "step": 820
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.056227684020996,
+      "learning_rate": 1.911642531732666e-05,
+      "loss": 0.5613,
+      "step": 830
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.2134612798690796,
+      "learning_rate": 1.907777478532909e-05,
+      "loss": 0.6439,
+      "step": 840
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 2.153582811355591,
+      "learning_rate": 1.9038337699485207e-05,
+      "loss": 0.6268,
+      "step": 850
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 1.4763509035110474,
+      "learning_rate": 1.8998117476864984e-05,
+      "loss": 0.6358,
+      "step": 860
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 2.120673656463623,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 0.5479,
+      "step": 870
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 3.2643983364105225,
+      "learning_rate": 1.8915341628552166e-05,
+      "loss": 0.5908,
+      "step": 880
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.7468228340148926,
+      "learning_rate": 1.8872793175064594e-05,
+      "loss": 0.6167,
+      "step": 890
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 1.6314669847488403,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 0.605,
+      "step": 900
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 1.2006853818893433,
+      "learning_rate": 1.8785393642396976e-05,
+      "loss": 0.6374,
+      "step": 910
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 5.40316915512085,
+      "learning_rate": 1.8740550136046195e-05,
+      "loss": 0.5919,
+      "step": 920
+    },
+    {
+      "epoch": 3.7199999999999998,
+      "grad_norm": 1.0310533046722412,
+      "learning_rate": 1.869494929505219e-05,
+      "loss": 0.596,
+      "step": 930
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 3.058582067489624,
+      "learning_rate": 1.8653264281300622e-05,
+      "loss": 0.5709,
+      "step": 940
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.8612210750579834,
+      "learning_rate": 1.8606235443821602e-05,
+      "loss": 0.6734,
+      "step": 950
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 1.0079221725463867,
+      "learning_rate": 1.8558460909544564e-05,
+      "loss": 0.5874,
+      "step": 960
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 1.8223471641540527,
+      "learning_rate": 1.850994481794692e-05,
+      "loss": 0.6199,
+      "step": 970
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 2.1497292518615723,
+      "learning_rate": 1.846069137275914e-05,
+      "loss": 0.5527,
+      "step": 980
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 1.5918537378311157,
+      "learning_rate": 1.8410704841600506e-05,
+      "loss": 0.5998,
+      "step": 990
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 4.21558952331543,
+      "learning_rate": 1.8359989555609355e-05,
+      "loss": 0.6357,
+      "step": 1000
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 2.512099504470825,
+      "learning_rate": 1.830854990906779e-05,
+      "loss": 0.6116,
+      "step": 1010
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 2.5760135650634766,
+      "learning_rate": 1.825639035902093e-05,
+      "loss": 0.54,
+      "step": 1020
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 1.5788276195526123,
+      "learning_rate": 1.8203515424890738e-05,
+      "loss": 0.6258,
+      "step": 1030
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 1.5123496055603027,
+      "learning_rate": 1.814992968808442e-05,
+      "loss": 0.5147,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 1.539919376373291,
+      "learning_rate": 1.809563779159746e-05,
+      "loss": 0.569,
+      "step": 1050
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.6823704242706299,
+      "learning_rate": 1.8040644439611348e-05,
+      "loss": 0.5588,
+      "step": 1060
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 1.1113232374191284,
+      "learning_rate": 1.798495439708594e-05,
+      "loss": 0.5692,
+      "step": 1070
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.6908786296844482,
+      "learning_rate": 1.792857248934663e-05,
+      "loss": 0.6102,
+      "step": 1080
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 1.7746518850326538,
+      "learning_rate": 1.7871503601666233e-05,
+      "loss": 0.5706,
+      "step": 1090
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.2888718843460083,
+      "learning_rate": 1.7813752678841702e-05,
+      "loss": 0.5964,
+      "step": 1100
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 2.7955427169799805,
+      "learning_rate": 1.7755324724765688e-05,
+      "loss": 0.6055,
+      "step": 1110
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.4672576189041138,
+      "learning_rate": 1.7696224801992947e-05,
+      "loss": 0.5548,
+      "step": 1120
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 2.2973852157592773,
+      "learning_rate": 1.7636458031301725e-05,
+      "loss": 0.5967,
+      "step": 1130
+    },
+    {
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.6734001636505127,
+      "learning_rate": 1.7576029591250036e-05,
+      "loss": 0.5567,
+      "step": 1140
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 2.127830743789673,
+      "learning_rate": 1.7514944717726962e-05,
+      "loss": 0.6065,
+      "step": 1150
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 2.201108455657959,
+      "learning_rate": 1.7453208703499006e-05,
+      "loss": 0.566,
+      "step": 1160
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 3.8374786376953125,
+      "learning_rate": 1.739082689775146e-05,
+      "loss": 0.55,
+      "step": 1170
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 2.7282190322875977,
+      "learning_rate": 1.732780470562496e-05,
+      "loss": 0.5748,
+      "step": 1180
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 1.8128880262374878,
+      "learning_rate": 1.7264147587747097e-05,
+      "loss": 0.6309,
+      "step": 1190
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 2.522096633911133,
+      "learning_rate": 1.7199861059759338e-05,
+      "loss": 0.5504,
+      "step": 1200
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 1.186700463294983,
+      "learning_rate": 1.7134950691839063e-05,
+      "loss": 0.5741,
+      "step": 1210
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 4.312258720397949,
+      "learning_rate": 1.7069422108216973e-05,
+      "loss": 0.575,
+      "step": 1220
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 3.402963399887085,
+      "learning_rate": 1.7003280986689733e-05,
+      "loss": 0.5842,
+      "step": 1230
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 2.346266269683838,
+      "learning_rate": 1.693653305812805e-05,
+      "loss": 0.5877,
+      "step": 1240
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.304891586303711,
+      "learning_rate": 1.686918410598009e-05,
+      "loss": 0.6153,
+      "step": 1250
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 2.874284505844116,
+      "learning_rate": 1.6801239965770366e-05,
+      "loss": 0.5776,
+      "step": 1260
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 3.4333293437957764,
+      "learning_rate": 1.6732706524594138e-05,
+      "loss": 0.5099,
+      "step": 1270
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 1.4809489250183105,
+      "learning_rate": 1.6663589720607287e-05,
+      "loss": 0.5535,
+      "step": 1280
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 3.005042314529419,
+      "learning_rate": 1.659389554251181e-05,
+      "loss": 0.561,
+      "step": 1290
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 1.8546274900436401,
+      "learning_rate": 1.652363002903693e-05,
+      "loss": 0.555,
+      "step": 1300
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 10.558731079101562,
+      "learning_rate": 1.6452799268415857e-05,
+      "loss": 0.5333,
+      "step": 1310
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 2.2776925563812256,
+      "learning_rate": 1.6381409397858257e-05,
+      "loss": 0.5941,
+      "step": 1320
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 2.6703994274139404,
+      "learning_rate": 1.6309466603018497e-05,
+      "loss": 0.5676,
+      "step": 1330
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 3.8042726516723633,
+      "learning_rate": 1.6236977117459693e-05,
+      "loss": 0.5609,
+      "step": 1340
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 1.2701845169067383,
+      "learning_rate": 1.616394722211357e-05,
+      "loss": 0.5702,
+      "step": 1350
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 1.392269253730774,
+      "learning_rate": 1.6090383244736256e-05,
+      "loss": 0.5388,
+      "step": 1360
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 2.634445905685425,
+      "learning_rate": 1.6016291559360023e-05,
+      "loss": 0.573,
+      "step": 1370
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 3.3800487518310547,
+      "learning_rate": 1.5941678585740976e-05,
+      "loss": 0.5522,
+      "step": 1380
+    },
+    {
+      "epoch": 5.5600000000000005,
+      "grad_norm": 2.935079336166382,
+      "learning_rate": 1.5866550788802815e-05,
+      "loss": 0.5615,
+      "step": 1390
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 1.9093431234359741,
+      "learning_rate": 1.579091467807668e-05,
+      "loss": 0.5537,
+      "step": 1400
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 2.825533628463745,
+      "learning_rate": 1.5714776807137128e-05,
+      "loss": 0.6197,
+      "step": 1410
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 3.172933578491211,
+      "learning_rate": 1.5638143773034268e-05,
+      "loss": 0.5415,
+      "step": 1420
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 1.2932432889938354,
+      "learning_rate": 1.556102221572219e-05,
+      "loss": 0.5792,
+      "step": 1430
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 2.1744470596313477,
+      "learning_rate": 1.5483418817483607e-05,
+      "loss": 0.52,
+      "step": 1440
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 2.4148924350738525,
+      "learning_rate": 1.540534030235087e-05,
+      "loss": 0.602,
+      "step": 1450
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 2.427771806716919,
+      "learning_rate": 1.5326793435523374e-05,
+      "loss": 0.515,
+      "step": 1460
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 1.566942811012268,
+      "learning_rate": 1.5247785022781343e-05,
+      "loss": 0.5795,
+      "step": 1470
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 1.7555649280548096,
+      "learning_rate": 1.5168321909896171e-05,
+      "loss": 0.5819,
+      "step": 1480
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 1.6367748975753784,
+      "learning_rate": 1.5088410982037251e-05,
+      "loss": 0.5244,
+      "step": 1490
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 1.9994490146636963,
+      "learning_rate": 1.50080591631754e-05,
+      "loss": 0.5823,
+      "step": 1500
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 6.0024261474609375,
+      "learning_rate": 1.4927273415482916e-05,
+      "loss": 0.5641,
+      "step": 1510
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 1.50034499168396,
+      "learning_rate": 1.484606073873035e-05,
+      "loss": 0.5325,
+      "step": 1520
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 12.008216857910156,
+      "learning_rate": 1.4764428169679987e-05,
+      "loss": 0.5384,
+      "step": 1530
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 1.6358847618103027,
+      "learning_rate": 1.4682382781476146e-05,
+      "loss": 0.5822,
+      "step": 1540
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 2.520883321762085,
+      "learning_rate": 1.4599931683032327e-05,
+      "loss": 0.5256,
+      "step": 1550
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 1.136460304260254,
+      "learning_rate": 1.4517082018415231e-05,
+      "loss": 0.5589,
+      "step": 1560
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 2.730435848236084,
+      "learning_rate": 1.4433840966225772e-05,
+      "loss": 0.4939,
+      "step": 1570
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.9575507044792175,
+      "learning_rate": 1.4350215738977077e-05,
+      "loss": 0.5277,
+      "step": 1580
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 1.9613964557647705,
+      "learning_rate": 1.4266213582469543e-05,
+      "loss": 0.5457,
+      "step": 1590
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 2.436429500579834,
+      "learning_rate": 1.4181841775163014e-05,
+      "loss": 0.547,
+      "step": 1600
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 2.1270251274108887,
+      "learning_rate": 1.409710762754615e-05,
+      "loss": 0.5422,
+      "step": 1610
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 2.4047000408172607,
+      "learning_rate": 1.4012018481502975e-05,
+      "loss": 0.5219,
+      "step": 1620
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 1.1183472871780396,
+      "learning_rate": 1.3926581709676752e-05,
+      "loss": 0.5088,
+      "step": 1630
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.89410400390625,
+      "learning_rate": 1.3840804714831164e-05,
+      "loss": 0.5707,
+      "step": 1640
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 1.2478140592575073,
+      "learning_rate": 1.3754694929208891e-05,
+      "loss": 0.5893,
+      "step": 1650
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 4.104971885681152,
+      "learning_rate": 1.3668259813887644e-05,
+      "loss": 0.5768,
+      "step": 1660
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 1.4798212051391602,
+      "learning_rate": 1.3581506858133677e-05,
+      "loss": 0.5223,
+      "step": 1670
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 3.716698408126831,
+      "learning_rate": 1.3494443578752893e-05,
+      "loss": 0.5208,
+      "step": 1680
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 2.887411117553711,
+      "learning_rate": 1.340707751943952e-05,
+      "loss": 0.5415,
+      "step": 1690
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 3.6842939853668213,
+      "learning_rate": 1.3319416250122484e-05,
+      "loss": 0.5272,
+      "step": 1700
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 4.174267292022705,
+      "learning_rate": 1.3231467366309523e-05,
+      "loss": 0.5341,
+      "step": 1710
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 4.511580467224121,
+      "learning_rate": 1.3143238488429042e-05,
+      "loss": 0.573,
+      "step": 1720
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 1.930474042892456,
+      "learning_rate": 1.3054737261169838e-05,
+      "loss": 0.5134,
+      "step": 1730
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 1.2770944833755493,
+      "learning_rate": 1.2965971352818736e-05,
+      "loss": 0.4917,
+      "step": 1740
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 2.182473659515381,
+      "learning_rate": 1.287694845459613e-05,
+      "loss": 0.5725,
+      "step": 1750
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 4.948633670806885,
+      "learning_rate": 1.2787676279989594e-05,
+      "loss": 0.5122,
+      "step": 1760
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 2.5241198539733887,
+      "learning_rate": 1.2698162564085536e-05,
+      "loss": 0.4839,
+      "step": 1770
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 1.8158693313598633,
+      "learning_rate": 1.2608415062898971e-05,
+      "loss": 0.4502,
+      "step": 1780
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 3.540010929107666,
+      "learning_rate": 1.2518441552701493e-05,
+      "loss": 0.5585,
+      "step": 1790
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 1.9738972187042236,
+      "learning_rate": 1.2428249829347509e-05,
+      "loss": 0.4918,
+      "step": 1800
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 1.0376015901565552,
+      "learning_rate": 1.2337847707598738e-05,
+      "loss": 0.4989,
+      "step": 1810
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 2.5017688274383545,
+      "learning_rate": 1.2247243020447104e-05,
+      "loss": 0.4962,
+      "step": 1820
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 1.2260102033615112,
+      "learning_rate": 1.2156443618436033e-05,
+      "loss": 0.5316,
+      "step": 1830
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 2.4500176906585693,
+      "learning_rate": 1.2065457368980236e-05,
+      "loss": 0.4841,
+      "step": 1840
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 3.0221171379089355,
+      "learning_rate": 1.197429215568403e-05,
+      "loss": 0.5507,
+      "step": 1850
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 5.26624059677124,
+      "learning_rate": 1.1882955877658252e-05,
+      "loss": 0.532,
+      "step": 1860
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 2.411428213119507,
+      "learning_rate": 1.1791456448835825e-05,
+      "loss": 0.4802,
+      "step": 1870
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 2.334620952606201,
+      "learning_rate": 1.169980179728606e-05,
+      "loss": 0.5331,
+      "step": 1880
+    },
+    {
+      "epoch": 7.5600000000000005,
+      "grad_norm": 2.4401047229766846,
+      "learning_rate": 1.1607999864527718e-05,
+      "loss": 0.4994,
+      "step": 1890
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 2.3867135047912598,
+      "learning_rate": 1.1516058604840891e-05,
+      "loss": 0.5124,
+      "step": 1900
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 2.3309555053710938,
+      "learning_rate": 1.1423985984577813e-05,
+      "loss": 0.574,
+      "step": 1910
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 1.1885383129119873,
+      "learning_rate": 1.1331789981472603e-05,
+      "loss": 0.5361,
+      "step": 1920
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 1.6586416959762573,
+      "learning_rate": 1.1239478583950019e-05,
+      "loss": 0.5388,
+      "step": 1930
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 1.3869335651397705,
+      "learning_rate": 1.1147059790433296e-05,
+      "loss": 0.536,
+      "step": 1940
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 1.5383076667785645,
+      "learning_rate": 1.1054541608651121e-05,
+      "loss": 0.5165,
+      "step": 1950
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 1.1627497673034668,
+      "learning_rate": 1.0961932054943778e-05,
+      "loss": 0.5369,
+      "step": 1960
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 1.4803476333618164,
+      "learning_rate": 1.0869239153568575e-05,
+      "loss": 0.548,
+      "step": 1970
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 1.503915786743164,
+      "learning_rate": 1.0776470936004572e-05,
+      "loss": 0.5377,
+      "step": 1980
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 1.9053574800491333,
+      "learning_rate": 1.0683635440256689e-05,
+      "loss": 0.5249,
+      "step": 1990
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 2.171719551086426,
+      "learning_rate": 1.059074071015923e-05,
+      "loss": 0.5162,
+      "step": 2000
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 3.7397103309631348,
+      "learning_rate": 1.0497794794678923e-05,
+      "loss": 0.5067,
+      "step": 2010
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 3.237569570541382,
+      "learning_rate": 1.0404805747217525e-05,
+      "loss": 0.4901,
+      "step": 2020
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 2.6131529808044434,
+      "learning_rate": 1.0311781624914e-05,
+      "loss": 0.4834,
+      "step": 2030
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 2.543020009994507,
+      "learning_rate": 1.0228036587536431e-05,
+      "loss": 0.4991,
+      "step": 2040
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 2.420510768890381,
+      "learning_rate": 1.013496803077246e-05,
+      "loss": 0.5326,
+      "step": 2050
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 1.7979626655578613,
+      "learning_rate": 1.0041887779554041e-05,
+      "loss": 0.501,
+      "step": 2060
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 3.0351650714874268,
+      "learning_rate": 9.948803898922586e-06,
+      "loss": 0.5263,
+      "step": 2070
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 2.1602799892425537,
+      "learning_rate": 9.85572445423399e-06,
+      "loss": 0.505,
+      "step": 2080
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 2.298388957977295,
+      "learning_rate": 9.762657510459784e-06,
+      "loss": 0.4962,
+      "step": 2090
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 1.9878581762313843,
+      "learning_rate": 9.669611131488346e-06,
+      "loss": 0.5086,
+      "step": 2100
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 3.1122074127197266,
+      "learning_rate": 9.576593379426196e-06,
+      "loss": 0.5105,
+      "step": 2110
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 1.8491990566253662,
+      "learning_rate": 9.483612313899436e-06,
+      "loss": 0.5028,
+      "step": 2120
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 2.2476413249969482,
+      "learning_rate": 9.390675991355435e-06,
+      "loss": 0.5273,
+      "step": 2130
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 3.6653342247009277,
+      "learning_rate": 9.297792464364748e-06,
+      "loss": 0.4313,
+      "step": 2140
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.8962536454200745,
+      "learning_rate": 9.204969780923404e-06,
+      "loss": 0.5045,
+      "step": 2150
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 6.7541823387146,
+      "learning_rate": 9.112215983755573e-06,
+      "loss": 0.4818,
+      "step": 2160
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 3.15523362159729,
+      "learning_rate": 9.019539109616694e-06,
+      "loss": 0.4779,
+      "step": 2170
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 1.048300862312317,
+      "learning_rate": 8.926947188597133e-06,
+      "loss": 0.4815,
+      "step": 2180
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 4.415710926055908,
+      "learning_rate": 8.8344482434264e-06,
+      "loss": 0.5259,
+      "step": 2190
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 4.474966049194336,
+      "learning_rate": 8.742050288778e-06,
+      "loss": 0.5378,
+      "step": 2200
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 3.487746477127075,
+      "learning_rate": 8.649761330575009e-06,
+      "loss": 0.5144,
+      "step": 2210
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 1.44117271900177,
+      "learning_rate": 8.557589365296385e-06,
+      "loss": 0.5383,
+      "step": 2220
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 2.9286913871765137,
+      "learning_rate": 8.4655423792841e-06,
+      "loss": 0.4653,
+      "step": 2230
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 1.4172818660736084,
+      "learning_rate": 8.373628348051165e-06,
+      "loss": 0.4868,
+      "step": 2240
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 1.9049030542373657,
+      "learning_rate": 8.281855235590574e-06,
+      "loss": 0.5606,
+      "step": 2250
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 3.6202874183654785,
+      "learning_rate": 8.19023099368526e-06,
+      "loss": 0.4717,
+      "step": 2260
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 1.4381736516952515,
+      "learning_rate": 8.098763561219101e-06,
+      "loss": 0.4578,
+      "step": 2270
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 3.8551642894744873,
+      "learning_rate": 8.007460863489042e-06,
+      "loss": 0.4553,
+      "step": 2280
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 2.2333943843841553,
+      "learning_rate": 7.91633081151841e-06,
+      "loss": 0.4861,
+      "step": 2290
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 3.517455816268921,
+      "learning_rate": 7.825381301371452e-06,
+      "loss": 0.4518,
+      "step": 2300
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 1.2912664413452148,
+      "learning_rate": 7.734620213469166e-06,
+      "loss": 0.4832,
+      "step": 2310
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 3.6948964595794678,
+      "learning_rate": 7.644055411906493e-06,
+      "loss": 0.4969,
+      "step": 2320
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 1.5953376293182373,
+      "learning_rate": 7.553694743770928e-06,
+      "loss": 0.4606,
+      "step": 2330
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 2.7939870357513428,
+      "learning_rate": 7.463546038462602e-06,
+      "loss": 0.5225,
+      "step": 2340
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 1.0297088623046875,
+      "learning_rate": 7.373617107015889e-06,
+      "loss": 0.529,
+      "step": 2350
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 2.87479305267334,
+      "learning_rate": 7.283915741422611e-06,
+      "loss": 0.5134,
+      "step": 2360
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 3.1623082160949707,
+      "learning_rate": 7.194449713956908e-06,
+      "loss": 0.4509,
+      "step": 2370
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 1.8917375802993774,
+      "learning_rate": 7.105226776501772e-06,
+      "loss": 0.5175,
+      "step": 2380
+    },
+    {
+      "epoch": 9.56,
+      "grad_norm": 1.6095237731933594,
+      "learning_rate": 7.016254659877398e-06,
+      "loss": 0.4742,
+      "step": 2390
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 3.2498207092285156,
+      "learning_rate": 6.927541073171333e-06,
+      "loss": 0.4605,
+      "step": 2400
+    },
+    {
+      "epoch": 9.64,
+      "grad_norm": 1.7395751476287842,
+      "learning_rate": 6.839093703070512e-06,
+      "loss": 0.4987,
+      "step": 2410
+    },
+    {
+      "epoch": 9.68,
+      "grad_norm": 2.4571480751037598,
+      "learning_rate": 6.750920213195238e-06,
+      "loss": 0.4829,
+      "step": 2420
+    },
+    {
+      "epoch": 9.72,
+      "grad_norm": 4.019631385803223,
+      "learning_rate": 6.6630282434351535e-06,
+      "loss": 0.4842,
+      "step": 2430
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 2.756540298461914,
+      "learning_rate": 6.575425409287292e-06,
+      "loss": 0.5198,
+      "step": 2440
+    },
+    {
+      "epoch": 9.8,
+      "grad_norm": 2.0040042400360107,
+      "learning_rate": 6.488119301196201e-06,
+      "loss": 0.5239,
+      "step": 2450
+    },
+    {
+      "epoch": 9.84,
+      "grad_norm": 7.419244766235352,
+      "learning_rate": 6.4011174838962706e-06,
+      "loss": 0.4636,
+      "step": 2460
+    },
+    {
+      "epoch": 9.88,
+      "grad_norm": 2.879230260848999,
+      "learning_rate": 6.314427495756283e-06,
+      "loss": 0.4693,
+      "step": 2470
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 2.1217892169952393,
+      "learning_rate": 6.228056848126236e-06,
+      "loss": 0.475,
+      "step": 2480
+    },
+    {
+      "epoch": 9.96,
+      "grad_norm": 2.1474809646606445,
+      "learning_rate": 6.142013024686509e-06,
+      "loss": 0.4995,
+      "step": 2490
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 4.3783721923828125,
+      "learning_rate": 6.056303480799449e-06,
+      "loss": 0.486,
+      "step": 2500
+    },
+    {
+      "epoch": 10.04,
+      "grad_norm": 1.397594928741455,
+      "learning_rate": 5.970935642863375e-06,
+      "loss": 0.4537,
+      "step": 2510
+    },
+    {
+      "epoch": 10.08,
+      "grad_norm": 3.2472903728485107,
+      "learning_rate": 5.885916907669114e-06,
+      "loss": 0.3856,
+      "step": 2520
+    },
+    {
+      "epoch": 10.12,
+      "grad_norm": 3.1067910194396973,
+      "learning_rate": 5.801254641759103e-06,
+      "loss": 0.4705,
+      "step": 2530
+    },
+    {
+      "epoch": 10.16,
+      "grad_norm": 2.5792055130004883,
+      "learning_rate": 5.716956180789098e-06,
+      "loss": 0.5011,
+      "step": 2540
+    },
+    {
+      "epoch": 10.2,
+      "grad_norm": 3.1249446868896484,
+      "learning_rate": 5.6330288288925805e-06,
+      "loss": 0.462,
+      "step": 2550
+    },
+    {
+      "epoch": 10.24,
+      "grad_norm": 4.708195209503174,
+      "learning_rate": 5.549479858047875e-06,
+      "loss": 0.5043,
+      "step": 2560
+    },
+    {
+      "epoch": 10.28,
+      "grad_norm": 1.4850634336471558,
+      "learning_rate": 5.466316507448049e-06,
+      "loss": 0.5244,
+      "step": 2570
+    },
+    {
+      "epoch": 10.32,
+      "grad_norm": 1.0298686027526855,
+      "learning_rate": 5.3835459828736945e-06,
+      "loss": 0.4362,
+      "step": 2580
+    },
+    {
+      "epoch": 10.36,
+      "grad_norm": 2.251105546951294,
+      "learning_rate": 5.30117545606854e-06,
+      "loss": 0.4788,
+      "step": 2590
+    },
+    {
+      "epoch": 10.4,
+      "grad_norm": 4.622702121734619,
+      "learning_rate": 5.219212064118079e-06,
+      "loss": 0.4265,
+      "step": 2600
+    },
+    {
+      "epoch": 10.44,
+      "grad_norm": 3.4977996349334717,
+      "learning_rate": 5.137662908831147e-06,
+      "loss": 0.5,
+      "step": 2610
+    },
+    {
+      "epoch": 10.48,
+      "grad_norm": 5.369349002838135,
+      "learning_rate": 5.056535056124592e-06,
+      "loss": 0.4409,
+      "step": 2620
+    },
+    {
+      "epoch": 10.52,
+      "grad_norm": 2.318140983581543,
+      "learning_rate": 4.97583553541102e-06,
+      "loss": 0.4594,
+      "step": 2630
+    },
+    {
+      "epoch": 10.56,
+      "grad_norm": 2.4116406440734863,
+      "learning_rate": 4.895571338989754e-06,
+      "loss": 0.4953,
+      "step": 2640
+    },
+    {
+      "epoch": 10.6,
+      "grad_norm": 3.0506629943847656,
+      "learning_rate": 4.8157494214409475e-06,
+      "loss": 0.4795,
+      "step": 2650
+    },
+    {
+      "epoch": 10.64,
+      "grad_norm": 1.8125630617141724,
+      "learning_rate": 4.736376699023023e-06,
+      "loss": 0.481,
+      "step": 2660
+    },
+    {
+      "epoch": 10.68,
+      "grad_norm": 1.9686360359191895,
+      "learning_rate": 4.6574600490733794e-06,
+      "loss": 0.4713,
+      "step": 2670
+    },
+    {
+      "epoch": 10.72,
+      "grad_norm": 2.8264060020446777,
+      "learning_rate": 4.579006309412533e-06,
+      "loss": 0.4501,
+      "step": 2680
+    },
+    {
+      "epoch": 10.76,
+      "grad_norm": 1.9835346937179565,
+      "learning_rate": 4.501022277751602e-06,
+      "loss": 0.4754,
+      "step": 2690
+    },
+    {
+      "epoch": 10.8,
+      "grad_norm": 4.487490653991699,
+      "learning_rate": 4.423514711103355e-06,
+      "loss": 0.5056,
+      "step": 2700
+    },
+    {
+      "epoch": 10.84,
+      "grad_norm": 3.1984522342681885,
+      "learning_rate": 4.346490325196704e-06,
+      "loss": 0.4415,
+      "step": 2710
+    },
+    {
+      "epoch": 10.88,
+      "grad_norm": 2.0367348194122314,
+      "learning_rate": 4.26995579389485e-06,
+      "loss": 0.5117,
+      "step": 2720
+    },
+    {
+      "epoch": 10.92,
+      "grad_norm": 1.78911292552948,
+      "learning_rate": 4.193917748616979e-06,
+      "loss": 0.475,
+      "step": 2730
+    },
+    {
+      "epoch": 10.96,
+      "grad_norm": 2.0589475631713867,
+      "learning_rate": 4.118382777763711e-06,
+      "loss": 0.4363,
+      "step": 2740
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 3.54664945602417,
+      "learning_rate": 4.04335742614622e-06,
+      "loss": 0.4665,
+      "step": 2750
+    },
+    {
+      "epoch": 11.04,
+      "grad_norm": 1.714920997619629,
+      "learning_rate": 3.968848194419163e-06,
+      "loss": 0.4515,
+      "step": 2760
+    },
+    {
+      "epoch": 11.08,
+      "grad_norm": 11.161652565002441,
+      "learning_rate": 3.894861538517401e-06,
+      "loss": 0.4285,
+      "step": 2770
+    },
+    {
+      "epoch": 11.12,
+      "grad_norm": 2.627831220626831,
+      "learning_rate": 3.821403869096658e-06,
+      "loss": 0.4343,
+      "step": 2780
+    },
+    {
+      "epoch": 11.16,
+      "grad_norm": 2.6865172386169434,
+      "learning_rate": 3.748481550978017e-06,
+      "loss": 0.4766,
+      "step": 2790
+    },
+    {
+      "epoch": 11.2,
+      "grad_norm": 4.996657848358154,
+      "learning_rate": 3.6761009025964657e-06,
+      "loss": 0.4096,
+      "step": 2800
+    },
+    {
+      "epoch": 11.24,
+      "grad_norm": 1.6282066106796265,
+      "learning_rate": 3.604268195453421e-06,
+      "loss": 0.4622,
+      "step": 2810
+    },
+    {
+      "epoch": 11.28,
+      "grad_norm": 1.6030402183532715,
+      "learning_rate": 3.5329896535733133e-06,
+      "loss": 0.4437,
+      "step": 2820
+    },
+    {
+      "epoch": 11.32,
+      "grad_norm": 2.8916399478912354,
+      "learning_rate": 3.462271452964321e-06,
+      "loss": 0.4871,
+      "step": 2830
+    },
+    {
+      "epoch": 11.36,
+      "grad_norm": 2.375190019607544,
+      "learning_rate": 3.3921197210832235e-06,
+      "loss": 0.4575,
+      "step": 2840
+    },
+    {
+      "epoch": 11.4,
+      "grad_norm": 4.021700382232666,
+      "learning_rate": 3.3225405363045016e-06,
+      "loss": 0.4699,
+      "step": 2850
+    },
+    {
+      "epoch": 11.44,
+      "grad_norm": 1.7844669818878174,
+      "learning_rate": 3.2535399273936407e-06,
+      "loss": 0.4648,
+      "step": 2860
+    },
+    {
+      "epoch": 11.48,
+      "grad_norm": 2.744528293609619,
+      "learning_rate": 3.1851238729848033e-06,
+      "loss": 0.3923,
+      "step": 2870
+    },
+    {
+      "epoch": 11.52,
+      "grad_norm": 1.6203703880310059,
+      "learning_rate": 3.11729830106276e-06,
+      "loss": 0.4717,
+      "step": 2880
+    },
+    {
+      "epoch": 11.56,
+      "grad_norm": 1.696370244026184,
+      "learning_rate": 3.0500690884492836e-06,
+      "loss": 0.4556,
+      "step": 2890
+    },
+    {
+      "epoch": 11.6,
+      "grad_norm": 4.04744291305542,
+      "learning_rate": 2.983442060293926e-06,
+      "loss": 0.4785,
+      "step": 2900
+    },
+    {
+      "epoch": 11.64,
+      "grad_norm": 2.629739284515381,
+      "learning_rate": 2.917422989569311e-06,
+      "loss": 0.463,
+      "step": 2910
+    },
+    {
+      "epoch": 11.68,
+      "grad_norm": 2.43945050239563,
+      "learning_rate": 2.852017596570901e-06,
+      "loss": 0.4551,
+      "step": 2920
+    },
+    {
+      "epoch": 11.72,
+      "grad_norm": 6.5788116455078125,
+      "learning_rate": 2.7872315484213954e-06,
+      "loss": 0.4501,
+      "step": 2930
+    },
+    {
+      "epoch": 11.76,
+      "grad_norm": 2.3283305168151855,
+      "learning_rate": 2.723070458579653e-06,
+      "loss": 0.4338,
+      "step": 2940
+    },
+    {
+      "epoch": 11.8,
+      "grad_norm": 4.168436527252197,
+      "learning_rate": 2.6595398863543407e-06,
+      "loss": 0.4744,
+      "step": 2950
+    },
+    {
+      "epoch": 11.84,
+      "grad_norm": 3.3579213619232178,
+      "learning_rate": 2.596645336422219e-06,
+      "loss": 0.4257,
+      "step": 2960
+    },
+    {
+      "epoch": 11.88,
+      "grad_norm": 4.208755970001221,
+      "learning_rate": 2.5343922583512026e-06,
+      "loss": 0.4676,
+      "step": 2970
+    },
+    {
+      "epoch": 11.92,
+      "grad_norm": 2.752279281616211,
+      "learning_rate": 2.472786046128156e-06,
+      "loss": 0.455,
+      "step": 2980
+    },
+    {
+      "epoch": 11.96,
+      "grad_norm": 3.5390079021453857,
+      "learning_rate": 2.411832037691545e-06,
+      "loss": 0.4646,
+      "step": 2990
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 2.805065870285034,
+      "learning_rate": 2.3515355144689155e-06,
+      "loss": 0.4774,
+      "step": 3000
+    },
+    {
+      "epoch": 12.04,
+      "grad_norm": 2.3203487396240234,
+      "learning_rate": 2.2919017009192703e-06,
+      "loss": 0.4333,
+      "step": 3010
+    },
+    {
+      "epoch": 12.08,
+      "grad_norm": 2.62695050239563,
+      "learning_rate": 2.2329357640804118e-06,
+      "loss": 0.456,
+      "step": 3020
+    },
+    {
+      "epoch": 12.12,
+      "grad_norm": 1.8128643035888672,
+      "learning_rate": 2.1746428131212126e-06,
+      "loss": 0.4054,
+      "step": 3030
+    },
+    {
+      "epoch": 12.16,
+      "grad_norm": 1.5641071796417236,
+      "learning_rate": 2.117027898898948e-06,
+      "loss": 0.4875,
+      "step": 3040
+    },
+    {
+      "epoch": 12.2,
+      "grad_norm": 2.1929521560668945,
+      "learning_rate": 2.0600960135216463e-06,
+      "loss": 0.4041,
+      "step": 3050
+    },
+    {
+      "epoch": 12.24,
+      "grad_norm": 1.746474027633667,
+      "learning_rate": 2.003852089915548e-06,
+      "loss": 0.5115,
+      "step": 3060
+    },
+    {
+      "epoch": 12.28,
+      "grad_norm": 2.731269121170044,
+      "learning_rate": 1.9483010013976766e-06,
+      "loss": 0.4459,
+      "step": 3070
+    },
+    {
+      "epoch": 12.32,
+      "grad_norm": 1.9771169424057007,
+      "learning_rate": 1.8934475612536019e-06,
+      "loss": 0.3677,
+      "step": 3080
+    },
+    {
+      "epoch": 12.36,
+      "grad_norm": 3.043891191482544,
+      "learning_rate": 1.8392965223203707e-06,
+      "loss": 0.4353,
+      "step": 3090
+    },
+    {
+      "epoch": 12.4,
+      "grad_norm": 3.361074447631836,
+      "learning_rate": 1.7858525765747047e-06,
+      "loss": 0.4578,
+      "step": 3100
+    },
+    {
+      "epoch": 12.44,
+      "grad_norm": 3.8681182861328125,
+      "learning_rate": 1.7331203547264452e-06,
+      "loss": 0.4057,
+      "step": 3110
+    },
+    {
+      "epoch": 12.48,
+      "grad_norm": 2.581637382507324,
+      "learning_rate": 1.6811044258173425e-06,
+      "loss": 0.4532,
+      "step": 3120
+    },
+    {
+      "epoch": 12.52,
+      "grad_norm": 2.8616292476654053,
+      "learning_rate": 1.629809296825139e-06,
+      "loss": 0.4551,
+      "step": 3130
+    },
+    {
+      "epoch": 12.56,
+      "grad_norm": 2.4617111682891846,
+      "learning_rate": 1.579239412273078e-06,
+      "loss": 0.4388,
+      "step": 3140
+    },
+    {
+      "epoch": 12.6,
+      "grad_norm": 14.222563743591309,
+      "learning_rate": 1.5293991538447882e-06,
+      "loss": 0.412,
+      "step": 3150
+    },
+    {
+      "epoch": 12.64,
+      "grad_norm": 3.5921924114227295,
+      "learning_rate": 1.4802928400046457e-06,
+      "loss": 0.4517,
+      "step": 3160
+    },
+    {
+      "epoch": 12.68,
+      "grad_norm": 2.4046990871429443,
+      "learning_rate": 1.4319247256235713e-06,
+      "loss": 0.4893,
+      "step": 3170
+    },
+    {
+      "epoch": 12.72,
+      "grad_norm": 2.5496039390563965,
+      "learning_rate": 1.3842990016103886e-06,
+      "loss": 0.4305,
+      "step": 3180
+    },
+    {
+      "epoch": 12.76,
+      "grad_norm": 2.3980159759521484,
+      "learning_rate": 1.3374197945486833e-06,
+      "loss": 0.3833,
+      "step": 3190
+    },
+    {
+      "epoch": 12.8,
+      "grad_norm": 1.515519142150879,
+      "learning_rate": 1.2912911663392468e-06,
+      "loss": 0.4513,
+      "step": 3200
+    },
+    {
+      "epoch": 12.84,
+      "grad_norm": 2.939988136291504,
+      "learning_rate": 1.245917113848144e-06,
+      "loss": 0.4712,
+      "step": 3210
+    },
+    {
+      "epoch": 12.88,
+      "grad_norm": 1.846997857093811,
+      "learning_rate": 1.2013015685603813e-06,
+      "loss": 0.4789,
+      "step": 3220
+    },
+    {
+      "epoch": 12.92,
+      "grad_norm": 2.960897445678711,
+      "learning_rate": 1.1574483962392768e-06,
+      "loss": 0.4128,
+      "step": 3230
+    },
+    {
+      "epoch": 12.96,
+      "grad_norm": 1.8638381958007812,
+      "learning_rate": 1.114361396591498e-06,
+      "loss": 0.4949,
+      "step": 3240
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 2.134097099304199,
+      "learning_rate": 1.0720443029378303e-06,
+      "loss": 0.4167,
+      "step": 3250
+    },
+    {
+      "epoch": 13.04,
+      "grad_norm": 2.2456820011138916,
+      "learning_rate": 1.0305007818897006e-06,
+      "loss": 0.4483,
+      "step": 3260
+    },
+    {
+      "epoch": 13.08,
+      "grad_norm": 9.045520782470703,
+      "learning_rate": 9.897344330314862e-07,
+      "loss": 0.454,
+      "step": 3270
+    },
+    {
+      "epoch": 13.12,
+      "grad_norm": 2.646930694580078,
+      "learning_rate": 9.497487886086132e-07,
+      "loss": 0.4438,
+      "step": 3280
+    },
+    {
+      "epoch": 13.16,
+      "grad_norm": 4.203260898590088,
+      "learning_rate": 9.105473132215126e-07,
+      "loss": 0.3904,
+      "step": 3290
+    },
+    {
+      "epoch": 13.2,
+      "grad_norm": 3.177109479904175,
+      "learning_rate": 8.721334035254203e-07,
+      "loss": 0.4128,
+      "step": 3300
+    },
+    {
+      "epoch": 13.24,
+      "grad_norm": 1.825671911239624,
+      "learning_rate": 8.345103879360695e-07,
+      "loss": 0.4479,
+      "step": 3310
+    },
+    {
+      "epoch": 13.28,
+      "grad_norm": 3.2267651557922363,
+      "learning_rate": 7.976815263412963e-07,
+      "loss": 0.3928,
+      "step": 3320
+    },
+    {
+      "epoch": 13.32,
+      "grad_norm": 3.6811180114746094,
+      "learning_rate": 7.616500098185908e-07,
+      "loss": 0.4163,
+      "step": 3330
+    },
+    {
+      "epoch": 13.36,
+      "grad_norm": 3.258467435836792,
+      "learning_rate": 7.264189603585892e-07,
+      "loss": 0.4186,
+      "step": 3340
+    },
+    {
+      "epoch": 13.4,
+      "grad_norm": 1.861440658569336,
+      "learning_rate": 6.919914305945774e-07,
+      "loss": 0.4416,
+      "step": 3350
+    },
+    {
+      "epoch": 13.44,
+      "grad_norm": 3.303765296936035,
+      "learning_rate": 6.58370403537989e-07,
+      "loss": 0.3958,
+      "step": 3360
+    },
+    {
+      "epoch": 13.48,
+      "grad_norm": 1.9865639209747314,
+      "learning_rate": 6.255587923199313e-07,
+      "loss": 0.4424,
+      "step": 3370
+    },
+    {
+      "epoch": 13.52,
+      "grad_norm": 2.3877573013305664,
+      "learning_rate": 5.935594399387856e-07,
+      "loss": 0.4778,
+      "step": 3380
+    },
+    {
+      "epoch": 13.56,
+      "grad_norm": 2.6566929817199707,
+      "learning_rate": 5.623751190138682e-07,
+      "loss": 0.4045,
+      "step": 3390
+    },
+    {
+      "epoch": 13.6,
+      "grad_norm": 4.0699052810668945,
+      "learning_rate": 5.320085315451862e-07,
+      "loss": 0.4275,
+      "step": 3400
+    },
+    {
+      "epoch": 13.64,
+      "grad_norm": 4.235281944274902,
+      "learning_rate": 5.024623086793323e-07,
+      "loss": 0.4346,
+      "step": 3410
+    },
+    {
+      "epoch": 13.68,
+      "grad_norm": 1.7936820983886719,
+      "learning_rate": 4.737390104814954e-07,
+      "loss": 0.4343,
+      "step": 3420
+    },
+    {
+      "epoch": 13.72,
+      "grad_norm": 4.244424343109131,
+      "learning_rate": 4.458411257136486e-07,
+      "loss": 0.4355,
+      "step": 3430
+    },
+    {
+      "epoch": 13.76,
+      "grad_norm": 3.0799527168273926,
+      "learning_rate": 4.1877107161890416e-07,
+      "loss": 0.4407,
+      "step": 3440
+    },
+    {
+      "epoch": 13.8,
+      "grad_norm": 3.4574971199035645,
+      "learning_rate": 3.9253119371206684e-07,
+      "loss": 0.4369,
+      "step": 3450
+    },
+    {
+      "epoch": 13.84,
+      "grad_norm": 1.3924965858459473,
+      "learning_rate": 3.671237655764104e-07,
+      "loss": 0.4823,
+      "step": 3460
+    },
+    {
+      "epoch": 13.88,
+      "grad_norm": 1.9487115144729614,
+      "learning_rate": 3.4255098866667114e-07,
+      "loss": 0.4566,
+      "step": 3470
+    },
+    {
+      "epoch": 13.92,
+      "grad_norm": 2.6502346992492676,
+      "learning_rate": 3.188149921183115e-07,
+      "loss": 0.4824,
+      "step": 3480
+    },
+    {
+      "epoch": 13.96,
+      "grad_norm": 3.1728081703186035,
+      "learning_rate": 2.959178325630296e-07,
+      "loss": 0.3983,
+      "step": 3490
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 2.273251533508301,
+      "learning_rate": 2.7386149395056463e-07,
+      "loss": 0.4541,
+      "step": 3500
+    },
+    {
+      "epoch": 14.04,
+      "grad_norm": 2.2681076526641846,
+      "learning_rate": 2.526478873767946e-07,
+      "loss": 0.4667,
+      "step": 3510
+    },
+    {
+      "epoch": 14.08,
+      "grad_norm": 2.3255577087402344,
+      "learning_rate": 2.322788509181484e-07,
+      "loss": 0.441,
+      "step": 3520
+    },
+    {
+      "epoch": 14.12,
+      "grad_norm": 1.8558521270751953,
+      "learning_rate": 2.1275614947233624e-07,
+      "loss": 0.4294,
+      "step": 3530
+    },
+    {
+      "epoch": 14.16,
+      "grad_norm": 3.000098943710327,
+      "learning_rate": 1.9408147460544203e-07,
+      "loss": 0.4245,
+      "step": 3540
+    },
+    {
+      "epoch": 14.2,
+      "grad_norm": 3.5599377155303955,
+      "learning_rate": 1.7625644440534384e-07,
+      "loss": 0.3884,
+      "step": 3550
+    },
+    {
+      "epoch": 14.24,
+      "grad_norm": 3.208889961242676,
+      "learning_rate": 1.5928260334151847e-07,
+      "loss": 0.4434,
+      "step": 3560
+    },
+    {
+      "epoch": 14.28,
+      "grad_norm": 3.4625139236450195,
+      "learning_rate": 1.4316142213121386e-07,
+      "loss": 0.4474,
+      "step": 3570
+    },
+    {
+      "epoch": 14.32,
+      "grad_norm": 4.883245944976807,
+      "learning_rate": 1.2789429761202565e-07,
+      "loss": 0.3927,
+      "step": 3580
+    },
+    {
+      "epoch": 14.36,
+      "grad_norm": 4.052615165710449,
+      "learning_rate": 1.134825526208605e-07,
+      "loss": 0.3967,
+      "step": 3590
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 1.4959968328475952,
+      "learning_rate": 9.992743587931674e-08,
+      "loss": 0.4807,
+      "step": 3600
+    },
+    {
+      "epoch": 14.44,
+      "grad_norm": 2.757499933242798,
+      "learning_rate": 8.723012188549318e-08,
+      "loss": 0.4506,
+      "step": 3610
+    },
+    {
+      "epoch": 14.48,
+      "grad_norm": 2.0996549129486084,
+      "learning_rate": 7.539171081221597e-08,
+      "loss": 0.3934,
+      "step": 3620
+    },
+    {
+      "epoch": 14.52,
+      "grad_norm": 2.4962105751037598,
+      "learning_rate": 6.44132284117216e-08,
+      "loss": 0.4549,
+      "step": 3630
+    },
+    {
+      "epoch": 14.56,
+      "grad_norm": 1.9287108182907104,
+      "learning_rate": 5.429562592677018e-08,
+      "loss": 0.4208,
+      "step": 3640
+    },
+    {
+      "epoch": 14.6,
+      "grad_norm": 4.647493362426758,
+      "learning_rate": 4.503978000823028e-08,
+      "loss": 0.4376,
+      "step": 3650
+    },
+    {
+      "epoch": 14.64,
+      "grad_norm": 1.24240243434906,
+      "learning_rate": 3.6646492639118567e-08,
+      "loss": 0.4562,
+      "step": 3660
+    },
+    {
+      "epoch": 14.68,
+      "grad_norm": 2.359651565551758,
+      "learning_rate": 2.911649106511316e-08,
+      "loss": 0.42,
+      "step": 3670
+    },
+    {
+      "epoch": 14.72,
+      "grad_norm": 2.7108898162841797,
+      "learning_rate": 2.2450427731534052e-08,
+      "loss": 0.4489,
+      "step": 3680
+    },
+    {
+      "epoch": 14.76,
+      "grad_norm": 1.4741981029510498,
+      "learning_rate": 1.664888022682165e-08,
+      "loss": 0.4532,
+      "step": 3690
+    },
+    {
+      "epoch": 14.8,
+      "grad_norm": 1.676648497581482,
+      "learning_rate": 1.1712351232480157e-08,
+      "loss": 0.4375,
+      "step": 3700
+    },
+    {
+      "epoch": 14.84,
+      "grad_norm": 1.2909196615219116,
+      "learning_rate": 7.641268479531283e-09,
+      "loss": 0.4574,
+      "step": 3710
+    },
+    {
+      "epoch": 14.88,
+      "grad_norm": 3.2380595207214355,
+      "learning_rate": 4.435984711446128e-09,
+      "loss": 0.3538,
+      "step": 3720
+    },
+    {
+      "epoch": 14.92,
+      "grad_norm": 1.9485658407211304,
+      "learning_rate": 2.0967776535851802e-09,
+      "loss": 0.3884,
+      "step": 3730
+    },
+    {
+      "epoch": 14.96,
+      "grad_norm": 2.0014994144439697,
+      "learning_rate": 6.238499891353389e-10,
+      "loss": 0.4266,
+      "step": 3740
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 2.8281970024108887,
+      "learning_rate": 1.7329341542859922e-11,
+      "loss": 0.4743,
+      "step": 3750
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 3750,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 15,
+  "save_steps": 10000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.815859931388314e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-3750/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7d13224a20770bb9bc99479f427296c451401a0aff0cc40e82e1b00254a2839
+size 5688

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c778a57a701c001dae45a84b62cab89df7f917e5066c9745e2952f4ada280a3
+oid sha256:26a8460fd5a0de85bee5542e09267a1fadbf54411155a8aa88b01bdbc7daf62d
 size 4486225984

runs/Feb22_18-06-29_cb38c9576c95/events.out.tfevents.1740247733.cb38c9576c95.125.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:218eb59f462f6b5bdf278485bb4faea5b7429ea5d9960fc4d325658623df92ac
+size 85373

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.0,
-    "total_flos": 2315448779538432.0,
-    "train_loss": 1.0368848323822022,
-    "train_runtime": 182.4958,
-    "train_samples_per_second": 1.096,
-    "train_steps_per_second": 0.548
+    "epoch": 15.0,
+    "total_flos": 8.815859931388314e+16,
+    "train_loss": 0.5605581335703532,
+    "train_runtime": 6084.0454,
+    "train_samples_per_second": 1.233,
+    "train_steps_per_second": 0.616
 }

trainer_log.jsonl

@@ -1,11 +1,376 @@
-{"current_steps": 10, "total_steps": 100, "loss": 1.5103, "lr": 2e-05, "epoch": 0.1, "percentage": 10.0, "elapsed_time": "0:00:16", "remaining_time": "0:02:24"}
-{"current_steps": 20, "total_steps": 100, "loss": 1.2272, "lr": 1.9396926207859085e-05, "epoch": 0.2, "percentage": 20.0, "elapsed_time": "0:00:31", "remaining_time": "0:02:05"}
-{"current_steps": 30, "total_steps": 100, "loss": 1.0918, "lr": 1.766044443118978e-05, "epoch": 0.3, "percentage": 30.0, "elapsed_time": "0:00:46", "remaining_time": "0:01:48"}
-{"current_steps": 40, "total_steps": 100, "loss": 1.0666, "lr": 1.5000000000000002e-05, "epoch": 0.4, "percentage": 40.0, "elapsed_time": "0:01:02", "remaining_time": "0:01:33"}
-{"current_steps": 50, "total_steps": 100, "loss": 0.9902, "lr": 1.1736481776669307e-05, "epoch": 0.5, "percentage": 50.0, "elapsed_time": "0:01:17", "remaining_time": "0:01:17"}
-{"current_steps": 60, "total_steps": 100, "loss": 0.8776, "lr": 8.263518223330698e-06, "epoch": 0.6, "percentage": 60.0, "elapsed_time": "0:01:34", "remaining_time": "0:01:02"}
-{"current_steps": 70, "total_steps": 100, "loss": 0.8724, "lr": 5.000000000000003e-06, "epoch": 0.7, "percentage": 70.0, "elapsed_time": "0:01:51", "remaining_time": "0:00:47"}
-{"current_steps": 80, "total_steps": 100, "loss": 0.935, "lr": 2.339555568810221e-06, "epoch": 0.8, "percentage": 80.0, "elapsed_time": "0:02:07", "remaining_time": "0:00:31"}
-{"current_steps": 90, "total_steps": 100, "loss": 0.9456, "lr": 6.030737921409169e-07, "epoch": 0.9, "percentage": 90.0, "elapsed_time": "0:02:25", "remaining_time": "0:00:16"}
-{"current_steps": 100, "total_steps": 100, "loss": 0.8522, "lr": 0.0, "epoch": 1.0, "percentage": 100.0, "elapsed_time": "0:02:40", "remaining_time": "0:00:00"}
-{"current_steps": 100, "total_steps": 100, "epoch": 1.0, "percentage": 100.0, "elapsed_time": "0:03:00", "remaining_time": "0:00:00"}
+{"current_steps": 10, "total_steps": 3750, "loss": 1.5494, "lr": 5.333333333333335e-07, "epoch": 0.04, "percentage": 0.27, "elapsed_time": "0:00:14", "remaining_time": "1:29:12"}
+{"current_steps": 20, "total_steps": 3750, "loss": 1.5781, "lr": 1.066666666666667e-06, "epoch": 0.08, "percentage": 0.53, "elapsed_time": "0:00:28", "remaining_time": "1:27:54"}
+{"current_steps": 30, "total_steps": 3750, "loss": 1.271, "lr": 1.6000000000000001e-06, "epoch": 0.12, "percentage": 0.8, "elapsed_time": "0:00:42", "remaining_time": "1:28:03"}
+{"current_steps": 40, "total_steps": 3750, "loss": 1.1263, "lr": 2.133333333333334e-06, "epoch": 0.16, "percentage": 1.07, "elapsed_time": "0:00:58", "remaining_time": "1:30:24"}
+{"current_steps": 50, "total_steps": 3750, "loss": 1.4759, "lr": 2.666666666666667e-06, "epoch": 0.2, "percentage": 1.33, "elapsed_time": "0:01:13", "remaining_time": "1:30:39"}
+{"current_steps": 60, "total_steps": 3750, "loss": 1.1906, "lr": 3.2000000000000003e-06, "epoch": 0.24, "percentage": 1.6, "elapsed_time": "0:01:30", "remaining_time": "1:32:38"}
+{"current_steps": 70, "total_steps": 3750, "loss": 1.3136, "lr": 3.7333333333333337e-06, "epoch": 0.28, "percentage": 1.87, "elapsed_time": "0:01:46", "remaining_time": "1:33:31"}
+{"current_steps": 80, "total_steps": 3750, "loss": 1.159, "lr": 4.266666666666668e-06, "epoch": 0.32, "percentage": 2.13, "elapsed_time": "0:02:03", "remaining_time": "1:34:08"}
+{"current_steps": 90, "total_steps": 3750, "loss": 1.0645, "lr": 4.800000000000001e-06, "epoch": 0.36, "percentage": 2.4, "elapsed_time": "0:02:19", "remaining_time": "1:34:32"}
+{"current_steps": 100, "total_steps": 3750, "loss": 1.3174, "lr": 5.333333333333334e-06, "epoch": 0.4, "percentage": 2.67, "elapsed_time": "0:02:34", "remaining_time": "1:34:03"}
+{"current_steps": 110, "total_steps": 3750, "loss": 1.0041, "lr": 5.8666666666666675e-06, "epoch": 0.44, "percentage": 2.93, "elapsed_time": "0:02:50", "remaining_time": "1:34:13"}
+{"current_steps": 120, "total_steps": 3750, "loss": 1.1208, "lr": 6.4000000000000006e-06, "epoch": 0.48, "percentage": 3.2, "elapsed_time": "0:03:06", "remaining_time": "1:34:11"}
+{"current_steps": 130, "total_steps": 3750, "loss": 0.9806, "lr": 6.9333333333333344e-06, "epoch": 0.52, "percentage": 3.47, "elapsed_time": "0:03:23", "remaining_time": "1:34:26"}
+{"current_steps": 140, "total_steps": 3750, "loss": 0.9904, "lr": 7.4666666666666675e-06, "epoch": 0.56, "percentage": 3.73, "elapsed_time": "0:03:39", "remaining_time": "1:34:17"}
+{"current_steps": 150, "total_steps": 3750, "loss": 0.9365, "lr": 8.000000000000001e-06, "epoch": 0.6, "percentage": 4.0, "elapsed_time": "0:03:55", "remaining_time": "1:34:00"}
+{"current_steps": 160, "total_steps": 3750, "loss": 0.8843, "lr": 8.533333333333335e-06, "epoch": 0.64, "percentage": 4.27, "elapsed_time": "0:04:10", "remaining_time": "1:33:44"}
+{"current_steps": 170, "total_steps": 3750, "loss": 0.9131, "lr": 9.066666666666667e-06, "epoch": 0.68, "percentage": 4.53, "elapsed_time": "0:04:27", "remaining_time": "1:33:47"}
+{"current_steps": 180, "total_steps": 3750, "loss": 0.9142, "lr": 9.600000000000001e-06, "epoch": 0.72, "percentage": 4.8, "elapsed_time": "0:04:42", "remaining_time": "1:33:32"}
+{"current_steps": 190, "total_steps": 3750, "loss": 0.8973, "lr": 1.0133333333333335e-05, "epoch": 0.76, "percentage": 5.07, "elapsed_time": "0:04:58", "remaining_time": "1:33:14"}
+{"current_steps": 200, "total_steps": 3750, "loss": 0.8407, "lr": 1.0666666666666667e-05, "epoch": 0.8, "percentage": 5.33, "elapsed_time": "0:05:14", "remaining_time": "1:33:02"}
+{"current_steps": 210, "total_steps": 3750, "loss": 0.8228, "lr": 1.1200000000000001e-05, "epoch": 0.84, "percentage": 5.6, "elapsed_time": "0:05:30", "remaining_time": "1:32:53"}
+{"current_steps": 220, "total_steps": 3750, "loss": 0.799, "lr": 1.1733333333333335e-05, "epoch": 0.88, "percentage": 5.87, "elapsed_time": "0:05:47", "remaining_time": "1:32:50"}
+{"current_steps": 230, "total_steps": 3750, "loss": 0.7759, "lr": 1.2266666666666667e-05, "epoch": 0.92, "percentage": 6.13, "elapsed_time": "0:06:03", "remaining_time": "1:32:40"}
+{"current_steps": 240, "total_steps": 3750, "loss": 0.7466, "lr": 1.2800000000000001e-05, "epoch": 0.96, "percentage": 6.4, "elapsed_time": "0:06:20", "remaining_time": "1:32:39"}
+{"current_steps": 250, "total_steps": 3750, "loss": 0.7317, "lr": 1.3333333333333333e-05, "epoch": 1.0, "percentage": 6.67, "elapsed_time": "0:06:36", "remaining_time": "1:32:33"}
+{"current_steps": 260, "total_steps": 3750, "loss": 0.7046, "lr": 1.3866666666666669e-05, "epoch": 1.04, "percentage": 6.93, "elapsed_time": "0:06:52", "remaining_time": "1:32:18"}
+{"current_steps": 270, "total_steps": 3750, "loss": 0.6625, "lr": 1.4400000000000001e-05, "epoch": 1.08, "percentage": 7.2, "elapsed_time": "0:07:08", "remaining_time": "1:32:03"}
+{"current_steps": 280, "total_steps": 3750, "loss": 0.7307, "lr": 1.4933333333333335e-05, "epoch": 1.12, "percentage": 7.47, "elapsed_time": "0:07:24", "remaining_time": "1:31:51"}
+{"current_steps": 290, "total_steps": 3750, "loss": 0.7996, "lr": 1.546666666666667e-05, "epoch": 1.16, "percentage": 7.73, "elapsed_time": "0:07:40", "remaining_time": "1:31:35"}
+{"current_steps": 300, "total_steps": 3750, "loss": 0.706, "lr": 1.6000000000000003e-05, "epoch": 1.2, "percentage": 8.0, "elapsed_time": "0:07:57", "remaining_time": "1:31:27"}
+{"current_steps": 310, "total_steps": 3750, "loss": 0.6941, "lr": 1.6533333333333333e-05, "epoch": 1.24, "percentage": 8.27, "elapsed_time": "0:08:13", "remaining_time": "1:31:20"}
+{"current_steps": 320, "total_steps": 3750, "loss": 0.746, "lr": 1.706666666666667e-05, "epoch": 1.28, "percentage": 8.53, "elapsed_time": "0:08:30", "remaining_time": "1:31:08"}
+{"current_steps": 330, "total_steps": 3750, "loss": 0.7262, "lr": 1.76e-05, "epoch": 1.32, "percentage": 8.8, "elapsed_time": "0:08:45", "remaining_time": "1:30:49"}
+{"current_steps": 340, "total_steps": 3750, "loss": 0.6552, "lr": 1.8133333333333335e-05, "epoch": 1.3599999999999999, "percentage": 9.07, "elapsed_time": "0:09:02", "remaining_time": "1:30:43"}
+{"current_steps": 350, "total_steps": 3750, "loss": 0.677, "lr": 1.866666666666667e-05, "epoch": 1.4, "percentage": 9.33, "elapsed_time": "0:09:18", "remaining_time": "1:30:27"}
+{"current_steps": 360, "total_steps": 3750, "loss": 0.6551, "lr": 1.9200000000000003e-05, "epoch": 1.44, "percentage": 9.6, "elapsed_time": "0:09:34", "remaining_time": "1:30:11"}
+{"current_steps": 370, "total_steps": 3750, "loss": 0.7076, "lr": 1.9733333333333336e-05, "epoch": 1.48, "percentage": 9.87, "elapsed_time": "0:09:50", "remaining_time": "1:29:57"}
+{"current_steps": 380, "total_steps": 3750, "loss": 0.6818, "lr": 1.999989169177959e-05, "epoch": 1.52, "percentage": 10.13, "elapsed_time": "0:10:06", "remaining_time": "1:29:43"}
+{"current_steps": 390, "total_steps": 3750, "loss": 0.6823, "lr": 1.9999025240093045e-05, "epoch": 1.56, "percentage": 10.4, "elapsed_time": "0:10:24", "remaining_time": "1:29:36"}
+{"current_steps": 400, "total_steps": 3750, "loss": 0.7704, "lr": 1.999729241179462e-05, "epoch": 1.6, "percentage": 10.67, "elapsed_time": "0:10:40", "remaining_time": "1:29:23"}
+{"current_steps": 410, "total_steps": 3750, "loss": 0.6668, "lr": 1.999469335702714e-05, "epoch": 1.6400000000000001, "percentage": 10.93, "elapsed_time": "0:10:56", "remaining_time": "1:29:08"}
+{"current_steps": 420, "total_steps": 3750, "loss": 0.6448, "lr": 1.9991228300988586e-05, "epoch": 1.6800000000000002, "percentage": 11.2, "elapsed_time": "0:11:12", "remaining_time": "1:28:51"}
+{"current_steps": 430, "total_steps": 3750, "loss": 0.7159, "lr": 1.998689754391257e-05, "epoch": 1.72, "percentage": 11.47, "elapsed_time": "0:11:29", "remaining_time": "1:28:39"}
+{"current_steps": 440, "total_steps": 3750, "loss": 0.6443, "lr": 1.998170146104234e-05, "epoch": 1.76, "percentage": 11.73, "elapsed_time": "0:11:44", "remaining_time": "1:28:20"}
+{"current_steps": 450, "total_steps": 3750, "loss": 0.7253, "lr": 1.9975640502598243e-05, "epoch": 1.8, "percentage": 12.0, "elapsed_time": "0:12:00", "remaining_time": "1:28:05"}
+{"current_steps": 460, "total_steps": 3750, "loss": 0.6349, "lr": 1.9968715193738738e-05, "epoch": 1.8399999999999999, "percentage": 12.27, "elapsed_time": "0:12:16", "remaining_time": "1:27:49"}
+{"current_steps": 470, "total_steps": 3750, "loss": 0.6793, "lr": 1.9960926134514875e-05, "epoch": 1.88, "percentage": 12.53, "elapsed_time": "0:12:32", "remaining_time": "1:27:31"}
+{"current_steps": 480, "total_steps": 3750, "loss": 0.6686, "lr": 1.9952273999818312e-05, "epoch": 1.92, "percentage": 12.8, "elapsed_time": "0:12:48", "remaining_time": "1:27:15"}
+{"current_steps": 490, "total_steps": 3750, "loss": 0.6584, "lr": 1.9942759539322845e-05, "epoch": 1.96, "percentage": 13.07, "elapsed_time": "0:13:04", "remaining_time": "1:27:01"}
+{"current_steps": 500, "total_steps": 3750, "loss": 0.6771, "lr": 1.9932383577419432e-05, "epoch": 2.0, "percentage": 13.33, "elapsed_time": "0:13:21", "remaining_time": "1:26:52"}
+{"current_steps": 510, "total_steps": 3750, "loss": 0.6664, "lr": 1.9921147013144782e-05, "epoch": 2.04, "percentage": 13.6, "elapsed_time": "0:13:39", "remaining_time": "1:26:43"}
+{"current_steps": 520, "total_steps": 3750, "loss": 0.6243, "lr": 1.990905082010344e-05, "epoch": 2.08, "percentage": 13.87, "elapsed_time": "0:13:55", "remaining_time": "1:26:27"}
+{"current_steps": 530, "total_steps": 3750, "loss": 0.6253, "lr": 1.9896096046383456e-05, "epoch": 2.12, "percentage": 14.13, "elapsed_time": "0:14:11", "remaining_time": "1:26:10"}
+{"current_steps": 540, "total_steps": 3750, "loss": 0.6362, "lr": 1.988228381446553e-05, "epoch": 2.16, "percentage": 14.4, "elapsed_time": "0:14:27", "remaining_time": "1:25:56"}
+{"current_steps": 550, "total_steps": 3750, "loss": 0.6517, "lr": 1.9867615321125796e-05, "epoch": 2.2, "percentage": 14.67, "elapsed_time": "0:14:44", "remaining_time": "1:25:44"}
+{"current_steps": 560, "total_steps": 3750, "loss": 0.6849, "lr": 1.985209183733209e-05, "epoch": 2.24, "percentage": 14.93, "elapsed_time": "0:15:00", "remaining_time": "1:25:27"}
+{"current_steps": 570, "total_steps": 3750, "loss": 0.6298, "lr": 1.983571470813386e-05, "epoch": 2.2800000000000002, "percentage": 15.2, "elapsed_time": "0:15:16", "remaining_time": "1:25:10"}
+{"current_steps": 580, "total_steps": 3750, "loss": 0.6588, "lr": 1.9818485352545595e-05, "epoch": 2.32, "percentage": 15.47, "elapsed_time": "0:15:32", "remaining_time": "1:24:57"}
+{"current_steps": 590, "total_steps": 3750, "loss": 0.6154, "lr": 1.980040526342388e-05, "epoch": 2.36, "percentage": 15.73, "elapsed_time": "0:15:49", "remaining_time": "1:24:42"}
+{"current_steps": 600, "total_steps": 3750, "loss": 0.6393, "lr": 1.9781476007338058e-05, "epoch": 2.4, "percentage": 16.0, "elapsed_time": "0:16:05", "remaining_time": "1:24:26"}
+{"current_steps": 610, "total_steps": 3750, "loss": 0.6245, "lr": 1.9761699224434476e-05, "epoch": 2.44, "percentage": 16.27, "elapsed_time": "0:16:20", "remaining_time": "1:24:08"}
+{"current_steps": 620, "total_steps": 3750, "loss": 0.6592, "lr": 1.9741076628294387e-05, "epoch": 2.48, "percentage": 16.53, "elapsed_time": "0:16:37", "remaining_time": "1:23:55"}
+{"current_steps": 630, "total_steps": 3750, "loss": 0.6262, "lr": 1.9719610005785466e-05, "epoch": 2.52, "percentage": 16.8, "elapsed_time": "0:16:53", "remaining_time": "1:23:39"}
+{"current_steps": 640, "total_steps": 3750, "loss": 0.672, "lr": 1.969730121690698e-05, "epoch": 2.56, "percentage": 17.07, "elapsed_time": "0:17:09", "remaining_time": "1:23:22"}
+{"current_steps": 650, "total_steps": 3750, "loss": 0.6057, "lr": 1.967415219462864e-05, "epoch": 2.6, "percentage": 17.33, "elapsed_time": "0:17:25", "remaining_time": "1:23:06"}
+{"current_steps": 660, "total_steps": 3750, "loss": 0.6142, "lr": 1.9650164944723116e-05, "epoch": 2.64, "percentage": 17.6, "elapsed_time": "0:17:41", "remaining_time": "1:22:51"}
+{"current_steps": 670, "total_steps": 3750, "loss": 0.6238, "lr": 1.9625341545592226e-05, "epoch": 2.68, "percentage": 17.87, "elapsed_time": "0:17:57", "remaining_time": "1:22:33"}
+{"current_steps": 680, "total_steps": 3750, "loss": 0.7166, "lr": 1.9599684148086876e-05, "epoch": 2.7199999999999998, "percentage": 18.13, "elapsed_time": "0:18:14", "remaining_time": "1:22:21"}
+{"current_steps": 690, "total_steps": 3750, "loss": 0.6769, "lr": 1.9573194975320672e-05, "epoch": 2.76, "percentage": 18.4, "elapsed_time": "0:18:30", "remaining_time": "1:22:03"}
+{"current_steps": 700, "total_steps": 3750, "loss": 0.6199, "lr": 1.954587632247732e-05, "epoch": 2.8, "percentage": 18.67, "elapsed_time": "0:18:46", "remaining_time": "1:21:47"}
+{"current_steps": 710, "total_steps": 3750, "loss": 0.6242, "lr": 1.951773055661174e-05, "epoch": 2.84, "percentage": 18.93, "elapsed_time": "0:19:02", "remaining_time": "1:21:30"}
+{"current_steps": 720, "total_steps": 3750, "loss": 0.6245, "lr": 1.9488760116444966e-05, "epoch": 2.88, "percentage": 19.2, "elapsed_time": "0:19:17", "remaining_time": "1:21:12"}
+{"current_steps": 730, "total_steps": 3750, "loss": 0.628, "lr": 1.9458967512152872e-05, "epoch": 2.92, "percentage": 19.47, "elapsed_time": "0:19:34", "remaining_time": "1:20:57"}
+{"current_steps": 740, "total_steps": 3750, "loss": 0.5806, "lr": 1.9428355325148632e-05, "epoch": 2.96, "percentage": 19.73, "elapsed_time": "0:19:49", "remaining_time": "1:20:39"}
+{"current_steps": 750, "total_steps": 3750, "loss": 0.5818, "lr": 1.9396926207859085e-05, "epoch": 3.0, "percentage": 20.0, "elapsed_time": "0:20:06", "remaining_time": "1:20:25"}
+{"current_steps": 760, "total_steps": 3750, "loss": 0.6387, "lr": 1.9364682883494892e-05, "epoch": 3.04, "percentage": 20.27, "elapsed_time": "0:20:23", "remaining_time": "1:20:12"}
+{"current_steps": 770, "total_steps": 3750, "loss": 0.6207, "lr": 1.9331628145814587e-05, "epoch": 3.08, "percentage": 20.53, "elapsed_time": "0:20:39", "remaining_time": "1:19:56"}
+{"current_steps": 780, "total_steps": 3750, "loss": 0.5868, "lr": 1.9297764858882516e-05, "epoch": 3.12, "percentage": 20.8, "elapsed_time": "0:20:54", "remaining_time": "1:19:38"}
+{"current_steps": 790, "total_steps": 3750, "loss": 0.5444, "lr": 1.926309595682066e-05, "epoch": 3.16, "percentage": 21.07, "elapsed_time": "0:21:10", "remaining_time": "1:19:21"}
+{"current_steps": 800, "total_steps": 3750, "loss": 0.5891, "lr": 1.9227624443554425e-05, "epoch": 3.2, "percentage": 21.33, "elapsed_time": "0:21:27", "remaining_time": "1:19:06"}
+{"current_steps": 810, "total_steps": 3750, "loss": 0.5453, "lr": 1.9191353392552346e-05, "epoch": 3.24, "percentage": 21.6, "elapsed_time": "0:21:43", "remaining_time": "1:18:51"}
+{"current_steps": 820, "total_steps": 3750, "loss": 0.6406, "lr": 1.9154285946559792e-05, "epoch": 3.2800000000000002, "percentage": 21.87, "elapsed_time": "0:21:59", "remaining_time": "1:18:36"}
+{"current_steps": 830, "total_steps": 3750, "loss": 0.5613, "lr": 1.911642531732666e-05, "epoch": 3.32, "percentage": 22.13, "elapsed_time": "0:22:15", "remaining_time": "1:18:19"}
+{"current_steps": 840, "total_steps": 3750, "loss": 0.6439, "lr": 1.907777478532909e-05, "epoch": 3.36, "percentage": 22.4, "elapsed_time": "0:22:32", "remaining_time": "1:18:04"}
+{"current_steps": 850, "total_steps": 3750, "loss": 0.6268, "lr": 1.9038337699485207e-05, "epoch": 3.4, "percentage": 22.67, "elapsed_time": "0:22:48", "remaining_time": "1:17:50"}
+{"current_steps": 860, "total_steps": 3750, "loss": 0.6358, "lr": 1.8998117476864984e-05, "epoch": 3.44, "percentage": 22.93, "elapsed_time": "0:23:06", "remaining_time": "1:17:37"}
+{"current_steps": 870, "total_steps": 3750, "loss": 0.5479, "lr": 1.895711760239413e-05, "epoch": 3.48, "percentage": 23.2, "elapsed_time": "0:23:21", "remaining_time": "1:17:21"}
+{"current_steps": 880, "total_steps": 3750, "loss": 0.5908, "lr": 1.8915341628552166e-05, "epoch": 3.52, "percentage": 23.47, "elapsed_time": "0:23:37", "remaining_time": "1:17:04"}
+{"current_steps": 890, "total_steps": 3750, "loss": 0.6167, "lr": 1.8872793175064594e-05, "epoch": 3.56, "percentage": 23.73, "elapsed_time": "0:23:53", "remaining_time": "1:16:46"}
+{"current_steps": 900, "total_steps": 3750, "loss": 0.605, "lr": 1.8829475928589272e-05, "epoch": 3.6, "percentage": 24.0, "elapsed_time": "0:24:08", "remaining_time": "1:16:27"}
+{"current_steps": 910, "total_steps": 3750, "loss": 0.6374, "lr": 1.8785393642396976e-05, "epoch": 3.64, "percentage": 24.27, "elapsed_time": "0:24:24", "remaining_time": "1:16:11"}
+{"current_steps": 920, "total_steps": 3750, "loss": 0.5919, "lr": 1.8740550136046195e-05, "epoch": 3.68, "percentage": 24.53, "elapsed_time": "0:24:40", "remaining_time": "1:15:52"}
+{"current_steps": 930, "total_steps": 3750, "loss": 0.596, "lr": 1.869494929505219e-05, "epoch": 3.7199999999999998, "percentage": 24.8, "elapsed_time": "0:24:55", "remaining_time": "1:15:36"}
+{"current_steps": 940, "total_steps": 3750, "loss": 0.5709, "lr": 1.8653264281300622e-05, "epoch": 3.76, "percentage": 25.07, "elapsed_time": "0:25:12", "remaining_time": "1:15:20"}
+{"current_steps": 950, "total_steps": 3750, "loss": 0.6734, "lr": 1.8606235443821602e-05, "epoch": 3.8, "percentage": 25.33, "elapsed_time": "0:25:28", "remaining_time": "1:15:05"}
+{"current_steps": 960, "total_steps": 3750, "loss": 0.5874, "lr": 1.8558460909544564e-05, "epoch": 3.84, "percentage": 25.6, "elapsed_time": "0:25:45", "remaining_time": "1:14:51"}
+{"current_steps": 970, "total_steps": 3750, "loss": 0.6199, "lr": 1.850994481794692e-05, "epoch": 3.88, "percentage": 25.87, "elapsed_time": "0:26:02", "remaining_time": "1:14:37"}
+{"current_steps": 980, "total_steps": 3750, "loss": 0.5527, "lr": 1.846069137275914e-05, "epoch": 3.92, "percentage": 26.13, "elapsed_time": "0:26:18", "remaining_time": "1:14:21"}
+{"current_steps": 990, "total_steps": 3750, "loss": 0.5998, "lr": 1.8410704841600506e-05, "epoch": 3.96, "percentage": 26.4, "elapsed_time": "0:26:34", "remaining_time": "1:14:04"}
+{"current_steps": 1000, "total_steps": 3750, "loss": 0.6357, "lr": 1.8359989555609355e-05, "epoch": 4.0, "percentage": 26.67, "elapsed_time": "0:26:49", "remaining_time": "1:13:47"}
+{"current_steps": 1010, "total_steps": 3750, "loss": 0.6116, "lr": 1.830854990906779e-05, "epoch": 4.04, "percentage": 26.93, "elapsed_time": "0:27:06", "remaining_time": "1:13:31"}
+{"current_steps": 1020, "total_steps": 3750, "loss": 0.54, "lr": 1.825639035902093e-05, "epoch": 4.08, "percentage": 27.2, "elapsed_time": "0:27:21", "remaining_time": "1:13:13"}
+{"current_steps": 1030, "total_steps": 3750, "loss": 0.6258, "lr": 1.8203515424890738e-05, "epoch": 4.12, "percentage": 27.47, "elapsed_time": "0:27:38", "remaining_time": "1:12:58"}
+{"current_steps": 1040, "total_steps": 3750, "loss": 0.5147, "lr": 1.814992968808442e-05, "epoch": 4.16, "percentage": 27.73, "elapsed_time": "0:27:53", "remaining_time": "1:12:41"}
+{"current_steps": 1050, "total_steps": 3750, "loss": 0.569, "lr": 1.809563779159746e-05, "epoch": 4.2, "percentage": 28.0, "elapsed_time": "0:28:10", "remaining_time": "1:12:26"}
+{"current_steps": 1060, "total_steps": 3750, "loss": 0.5588, "lr": 1.8040644439611348e-05, "epoch": 4.24, "percentage": 28.27, "elapsed_time": "0:28:26", "remaining_time": "1:12:09"}
+{"current_steps": 1070, "total_steps": 3750, "loss": 0.5692, "lr": 1.798495439708594e-05, "epoch": 4.28, "percentage": 28.53, "elapsed_time": "0:28:41", "remaining_time": "1:11:52"}
+{"current_steps": 1080, "total_steps": 3750, "loss": 0.6102, "lr": 1.792857248934663e-05, "epoch": 4.32, "percentage": 28.8, "elapsed_time": "0:28:57", "remaining_time": "1:11:36"}
+{"current_steps": 1090, "total_steps": 3750, "loss": 0.5706, "lr": 1.7871503601666233e-05, "epoch": 4.36, "percentage": 29.07, "elapsed_time": "0:29:13", "remaining_time": "1:11:19"}
+{"current_steps": 1100, "total_steps": 3750, "loss": 0.5964, "lr": 1.7813752678841702e-05, "epoch": 4.4, "percentage": 29.33, "elapsed_time": "0:29:29", "remaining_time": "1:11:03"}
+{"current_steps": 1110, "total_steps": 3750, "loss": 0.6055, "lr": 1.7755324724765688e-05, "epoch": 4.44, "percentage": 29.6, "elapsed_time": "0:29:45", "remaining_time": "1:10:47"}
+{"current_steps": 1120, "total_steps": 3750, "loss": 0.5548, "lr": 1.7696224801992947e-05, "epoch": 4.48, "percentage": 29.87, "elapsed_time": "0:30:02", "remaining_time": "1:10:31"}
+{"current_steps": 1130, "total_steps": 3750, "loss": 0.5967, "lr": 1.7636458031301725e-05, "epoch": 4.52, "percentage": 30.13, "elapsed_time": "0:30:19", "remaining_time": "1:10:18"}
+{"current_steps": 1140, "total_steps": 3750, "loss": 0.5567, "lr": 1.7576029591250036e-05, "epoch": 4.5600000000000005, "percentage": 30.4, "elapsed_time": "0:30:35", "remaining_time": "1:10:02"}
+{"current_steps": 1150, "total_steps": 3750, "loss": 0.6065, "lr": 1.7514944717726962e-05, "epoch": 4.6, "percentage": 30.67, "elapsed_time": "0:30:51", "remaining_time": "1:09:46"}
+{"current_steps": 1160, "total_steps": 3750, "loss": 0.566, "lr": 1.7453208703499006e-05, "epoch": 4.64, "percentage": 30.93, "elapsed_time": "0:31:07", "remaining_time": "1:09:29"}
+{"current_steps": 1170, "total_steps": 3750, "loss": 0.55, "lr": 1.739082689775146e-05, "epoch": 4.68, "percentage": 31.2, "elapsed_time": "0:31:24", "remaining_time": "1:09:15"}
+{"current_steps": 1180, "total_steps": 3750, "loss": 0.5748, "lr": 1.732780470562496e-05, "epoch": 4.72, "percentage": 31.47, "elapsed_time": "0:31:40", "remaining_time": "1:08:59"}
+{"current_steps": 1190, "total_steps": 3750, "loss": 0.6309, "lr": 1.7264147587747097e-05, "epoch": 4.76, "percentage": 31.73, "elapsed_time": "0:31:57", "remaining_time": "1:08:45"}
+{"current_steps": 1200, "total_steps": 3750, "loss": 0.5504, "lr": 1.7199861059759338e-05, "epoch": 4.8, "percentage": 32.0, "elapsed_time": "0:32:13", "remaining_time": "1:08:28"}
+{"current_steps": 1210, "total_steps": 3750, "loss": 0.5741, "lr": 1.7134950691839063e-05, "epoch": 4.84, "percentage": 32.27, "elapsed_time": "0:32:28", "remaining_time": "1:08:11"}
+{"current_steps": 1220, "total_steps": 3750, "loss": 0.575, "lr": 1.7069422108216973e-05, "epoch": 4.88, "percentage": 32.53, "elapsed_time": "0:32:45", "remaining_time": "1:07:56"}
+{"current_steps": 1230, "total_steps": 3750, "loss": 0.5842, "lr": 1.7003280986689733e-05, "epoch": 4.92, "percentage": 32.8, "elapsed_time": "0:33:01", "remaining_time": "1:07:40"}
+{"current_steps": 1240, "total_steps": 3750, "loss": 0.5877, "lr": 1.693653305812805e-05, "epoch": 4.96, "percentage": 33.07, "elapsed_time": "0:33:18", "remaining_time": "1:07:24"}
+{"current_steps": 1250, "total_steps": 3750, "loss": 0.6153, "lr": 1.686918410598009e-05, "epoch": 5.0, "percentage": 33.33, "elapsed_time": "0:33:34", "remaining_time": "1:07:09"}
+{"current_steps": 1260, "total_steps": 3750, "loss": 0.5776, "lr": 1.6801239965770366e-05, "epoch": 5.04, "percentage": 33.6, "elapsed_time": "0:33:51", "remaining_time": "1:06:54"}
+{"current_steps": 1270, "total_steps": 3750, "loss": 0.5099, "lr": 1.6732706524594138e-05, "epoch": 5.08, "percentage": 33.87, "elapsed_time": "0:34:06", "remaining_time": "1:06:37"}
+{"current_steps": 1280, "total_steps": 3750, "loss": 0.5535, "lr": 1.6663589720607287e-05, "epoch": 5.12, "percentage": 34.13, "elapsed_time": "0:34:22", "remaining_time": "1:06:20"}
+{"current_steps": 1290, "total_steps": 3750, "loss": 0.561, "lr": 1.659389554251181e-05, "epoch": 5.16, "percentage": 34.4, "elapsed_time": "0:34:39", "remaining_time": "1:06:06"}
+{"current_steps": 1300, "total_steps": 3750, "loss": 0.555, "lr": 1.652363002903693e-05, "epoch": 5.2, "percentage": 34.67, "elapsed_time": "0:34:55", "remaining_time": "1:05:49"}
+{"current_steps": 1310, "total_steps": 3750, "loss": 0.5333, "lr": 1.6452799268415857e-05, "epoch": 5.24, "percentage": 34.93, "elapsed_time": "0:35:11", "remaining_time": "1:05:33"}
+{"current_steps": 1320, "total_steps": 3750, "loss": 0.5941, "lr": 1.6381409397858257e-05, "epoch": 5.28, "percentage": 35.2, "elapsed_time": "0:35:28", "remaining_time": "1:05:17"}
+{"current_steps": 1330, "total_steps": 3750, "loss": 0.5676, "lr": 1.6309466603018497e-05, "epoch": 5.32, "percentage": 35.47, "elapsed_time": "0:35:44", "remaining_time": "1:05:01"}
+{"current_steps": 1340, "total_steps": 3750, "loss": 0.5609, "lr": 1.6236977117459693e-05, "epoch": 5.36, "percentage": 35.73, "elapsed_time": "0:35:59", "remaining_time": "1:04:44"}
+{"current_steps": 1350, "total_steps": 3750, "loss": 0.5702, "lr": 1.616394722211357e-05, "epoch": 5.4, "percentage": 36.0, "elapsed_time": "0:36:16", "remaining_time": "1:04:28"}
+{"current_steps": 1360, "total_steps": 3750, "loss": 0.5388, "lr": 1.6090383244736256e-05, "epoch": 5.44, "percentage": 36.27, "elapsed_time": "0:36:32", "remaining_time": "1:04:13"}
+{"current_steps": 1370, "total_steps": 3750, "loss": 0.573, "lr": 1.6016291559360023e-05, "epoch": 5.48, "percentage": 36.53, "elapsed_time": "0:36:48", "remaining_time": "1:03:57"}
+{"current_steps": 1380, "total_steps": 3750, "loss": 0.5522, "lr": 1.5941678585740976e-05, "epoch": 5.52, "percentage": 36.8, "elapsed_time": "0:37:04", "remaining_time": "1:03:40"}
+{"current_steps": 1390, "total_steps": 3750, "loss": 0.5615, "lr": 1.5866550788802815e-05, "epoch": 5.5600000000000005, "percentage": 37.07, "elapsed_time": "0:37:21", "remaining_time": "1:03:25"}
+{"current_steps": 1400, "total_steps": 3750, "loss": 0.5537, "lr": 1.579091467807668e-05, "epoch": 5.6, "percentage": 37.33, "elapsed_time": "0:37:37", "remaining_time": "1:03:09"}
+{"current_steps": 1410, "total_steps": 3750, "loss": 0.6197, "lr": 1.5714776807137128e-05, "epoch": 5.64, "percentage": 37.6, "elapsed_time": "0:37:53", "remaining_time": "1:02:53"}
+{"current_steps": 1420, "total_steps": 3750, "loss": 0.5415, "lr": 1.5638143773034268e-05, "epoch": 5.68, "percentage": 37.87, "elapsed_time": "0:38:09", "remaining_time": "1:02:36"}
+{"current_steps": 1430, "total_steps": 3750, "loss": 0.5792, "lr": 1.556102221572219e-05, "epoch": 5.72, "percentage": 38.13, "elapsed_time": "0:38:25", "remaining_time": "1:02:21"}
+{"current_steps": 1440, "total_steps": 3750, "loss": 0.52, "lr": 1.5483418817483607e-05, "epoch": 5.76, "percentage": 38.4, "elapsed_time": "0:38:41", "remaining_time": "1:02:04"}
+{"current_steps": 1450, "total_steps": 3750, "loss": 0.602, "lr": 1.540534030235087e-05, "epoch": 5.8, "percentage": 38.67, "elapsed_time": "0:38:58", "remaining_time": "1:01:49"}
+{"current_steps": 1460, "total_steps": 3750, "loss": 0.515, "lr": 1.5326793435523374e-05, "epoch": 5.84, "percentage": 38.93, "elapsed_time": "0:39:14", "remaining_time": "1:01:33"}
+{"current_steps": 1470, "total_steps": 3750, "loss": 0.5795, "lr": 1.5247785022781343e-05, "epoch": 5.88, "percentage": 39.2, "elapsed_time": "0:39:30", "remaining_time": "1:01:17"}
+{"current_steps": 1480, "total_steps": 3750, "loss": 0.5819, "lr": 1.5168321909896171e-05, "epoch": 5.92, "percentage": 39.47, "elapsed_time": "0:39:47", "remaining_time": "1:01:01"}
+{"current_steps": 1490, "total_steps": 3750, "loss": 0.5244, "lr": 1.5088410982037251e-05, "epoch": 5.96, "percentage": 39.73, "elapsed_time": "0:40:03", "remaining_time": "1:00:45"}
+{"current_steps": 1500, "total_steps": 3750, "loss": 0.5823, "lr": 1.50080591631754e-05, "epoch": 6.0, "percentage": 40.0, "elapsed_time": "0:40:19", "remaining_time": "1:00:29"}
+{"current_steps": 1510, "total_steps": 3750, "loss": 0.5641, "lr": 1.4927273415482916e-05, "epoch": 6.04, "percentage": 40.27, "elapsed_time": "0:40:36", "remaining_time": "1:00:13"}
+{"current_steps": 1520, "total_steps": 3750, "loss": 0.5325, "lr": 1.484606073873035e-05, "epoch": 6.08, "percentage": 40.53, "elapsed_time": "0:40:51", "remaining_time": "0:59:57"}
+{"current_steps": 1530, "total_steps": 3750, "loss": 0.5384, "lr": 1.4764428169679987e-05, "epoch": 6.12, "percentage": 40.8, "elapsed_time": "0:41:08", "remaining_time": "0:59:41"}
+{"current_steps": 1540, "total_steps": 3750, "loss": 0.5822, "lr": 1.4682382781476146e-05, "epoch": 6.16, "percentage": 41.07, "elapsed_time": "0:41:25", "remaining_time": "0:59:26"}
+{"current_steps": 1550, "total_steps": 3750, "loss": 0.5256, "lr": 1.4599931683032327e-05, "epoch": 6.2, "percentage": 41.33, "elapsed_time": "0:41:41", "remaining_time": "0:59:10"}
+{"current_steps": 1560, "total_steps": 3750, "loss": 0.5589, "lr": 1.4517082018415231e-05, "epoch": 6.24, "percentage": 41.6, "elapsed_time": "0:41:58", "remaining_time": "0:58:55"}
+{"current_steps": 1570, "total_steps": 3750, "loss": 0.4939, "lr": 1.4433840966225772e-05, "epoch": 6.28, "percentage": 41.87, "elapsed_time": "0:42:14", "remaining_time": "0:58:39"}
+{"current_steps": 1580, "total_steps": 3750, "loss": 0.5277, "lr": 1.4350215738977077e-05, "epoch": 6.32, "percentage": 42.13, "elapsed_time": "0:42:30", "remaining_time": "0:58:23"}
+{"current_steps": 1590, "total_steps": 3750, "loss": 0.5457, "lr": 1.4266213582469543e-05, "epoch": 6.36, "percentage": 42.4, "elapsed_time": "0:42:46", "remaining_time": "0:58:07"}
+{"current_steps": 1600, "total_steps": 3750, "loss": 0.547, "lr": 1.4181841775163014e-05, "epoch": 6.4, "percentage": 42.67, "elapsed_time": "0:43:03", "remaining_time": "0:57:51"}
+{"current_steps": 1610, "total_steps": 3750, "loss": 0.5422, "lr": 1.409710762754615e-05, "epoch": 6.44, "percentage": 42.93, "elapsed_time": "0:43:19", "remaining_time": "0:57:35"}
+{"current_steps": 1620, "total_steps": 3750, "loss": 0.5219, "lr": 1.4012018481502975e-05, "epoch": 6.48, "percentage": 43.2, "elapsed_time": "0:43:34", "remaining_time": "0:57:17"}
+{"current_steps": 1630, "total_steps": 3750, "loss": 0.5088, "lr": 1.3926581709676752e-05, "epoch": 6.52, "percentage": 43.47, "elapsed_time": "0:43:51", "remaining_time": "0:57:02"}
+{"current_steps": 1640, "total_steps": 3750, "loss": 0.5707, "lr": 1.3840804714831164e-05, "epoch": 6.5600000000000005, "percentage": 43.73, "elapsed_time": "0:44:07", "remaining_time": "0:56:46"}
+{"current_steps": 1650, "total_steps": 3750, "loss": 0.5893, "lr": 1.3754694929208891e-05, "epoch": 6.6, "percentage": 44.0, "elapsed_time": "0:44:24", "remaining_time": "0:56:31"}
+{"current_steps": 1660, "total_steps": 3750, "loss": 0.5768, "lr": 1.3668259813887644e-05, "epoch": 6.64, "percentage": 44.27, "elapsed_time": "0:44:41", "remaining_time": "0:56:15"}
+{"current_steps": 1670, "total_steps": 3750, "loss": 0.5223, "lr": 1.3581506858133677e-05, "epoch": 6.68, "percentage": 44.53, "elapsed_time": "0:44:57", "remaining_time": "0:55:59"}
+{"current_steps": 1680, "total_steps": 3750, "loss": 0.5208, "lr": 1.3494443578752893e-05, "epoch": 6.72, "percentage": 44.8, "elapsed_time": "0:45:13", "remaining_time": "0:55:42"}
+{"current_steps": 1690, "total_steps": 3750, "loss": 0.5415, "lr": 1.340707751943952e-05, "epoch": 6.76, "percentage": 45.07, "elapsed_time": "0:45:29", "remaining_time": "0:55:26"}
+{"current_steps": 1700, "total_steps": 3750, "loss": 0.5272, "lr": 1.3319416250122484e-05, "epoch": 6.8, "percentage": 45.33, "elapsed_time": "0:45:45", "remaining_time": "0:55:10"}
+{"current_steps": 1710, "total_steps": 3750, "loss": 0.5341, "lr": 1.3231467366309523e-05, "epoch": 6.84, "percentage": 45.6, "elapsed_time": "0:46:01", "remaining_time": "0:54:54"}
+{"current_steps": 1720, "total_steps": 3750, "loss": 0.573, "lr": 1.3143238488429042e-05, "epoch": 6.88, "percentage": 45.87, "elapsed_time": "0:46:17", "remaining_time": "0:54:38"}
+{"current_steps": 1730, "total_steps": 3750, "loss": 0.5134, "lr": 1.3054737261169838e-05, "epoch": 6.92, "percentage": 46.13, "elapsed_time": "0:46:33", "remaining_time": "0:54:21"}
+{"current_steps": 1740, "total_steps": 3750, "loss": 0.4917, "lr": 1.2965971352818736e-05, "epoch": 6.96, "percentage": 46.4, "elapsed_time": "0:46:49", "remaining_time": "0:54:05"}
+{"current_steps": 1750, "total_steps": 3750, "loss": 0.5725, "lr": 1.287694845459613e-05, "epoch": 7.0, "percentage": 46.67, "elapsed_time": "0:47:05", "remaining_time": "0:53:48"}
+{"current_steps": 1760, "total_steps": 3750, "loss": 0.5122, "lr": 1.2787676279989594e-05, "epoch": 7.04, "percentage": 46.93, "elapsed_time": "0:47:21", "remaining_time": "0:53:32"}
+{"current_steps": 1770, "total_steps": 3750, "loss": 0.4839, "lr": 1.2698162564085536e-05, "epoch": 7.08, "percentage": 47.2, "elapsed_time": "0:47:37", "remaining_time": "0:53:16"}
+{"current_steps": 1780, "total_steps": 3750, "loss": 0.4502, "lr": 1.2608415062898971e-05, "epoch": 7.12, "percentage": 47.47, "elapsed_time": "0:47:52", "remaining_time": "0:52:59"}
+{"current_steps": 1790, "total_steps": 3750, "loss": 0.5585, "lr": 1.2518441552701493e-05, "epoch": 7.16, "percentage": 47.73, "elapsed_time": "0:48:08", "remaining_time": "0:52:43"}
+{"current_steps": 1800, "total_steps": 3750, "loss": 0.4918, "lr": 1.2428249829347509e-05, "epoch": 7.2, "percentage": 48.0, "elapsed_time": "0:48:25", "remaining_time": "0:52:27"}
+{"current_steps": 1810, "total_steps": 3750, "loss": 0.4989, "lr": 1.2337847707598738e-05, "epoch": 7.24, "percentage": 48.27, "elapsed_time": "0:48:41", "remaining_time": "0:52:11"}
+{"current_steps": 1820, "total_steps": 3750, "loss": 0.4962, "lr": 1.2247243020447104e-05, "epoch": 7.28, "percentage": 48.53, "elapsed_time": "0:48:57", "remaining_time": "0:51:55"}
+{"current_steps": 1830, "total_steps": 3750, "loss": 0.5316, "lr": 1.2156443618436033e-05, "epoch": 7.32, "percentage": 48.8, "elapsed_time": "0:49:14", "remaining_time": "0:51:39"}
+{"current_steps": 1840, "total_steps": 3750, "loss": 0.4841, "lr": 1.2065457368980236e-05, "epoch": 7.36, "percentage": 49.07, "elapsed_time": "0:49:29", "remaining_time": "0:51:22"}
+{"current_steps": 1850, "total_steps": 3750, "loss": 0.5507, "lr": 1.197429215568403e-05, "epoch": 7.4, "percentage": 49.33, "elapsed_time": "0:49:46", "remaining_time": "0:51:07"}
+{"current_steps": 1860, "total_steps": 3750, "loss": 0.532, "lr": 1.1882955877658252e-05, "epoch": 7.44, "percentage": 49.6, "elapsed_time": "0:50:02", "remaining_time": "0:50:51"}
+{"current_steps": 1870, "total_steps": 3750, "loss": 0.4802, "lr": 1.1791456448835825e-05, "epoch": 7.48, "percentage": 49.87, "elapsed_time": "0:50:18", "remaining_time": "0:50:34"}
+{"current_steps": 1880, "total_steps": 3750, "loss": 0.5331, "lr": 1.169980179728606e-05, "epoch": 7.52, "percentage": 50.13, "elapsed_time": "0:50:34", "remaining_time": "0:50:18"}
+{"current_steps": 1890, "total_steps": 3750, "loss": 0.4994, "lr": 1.1607999864527718e-05, "epoch": 7.5600000000000005, "percentage": 50.4, "elapsed_time": "0:50:49", "remaining_time": "0:50:01"}
+{"current_steps": 1900, "total_steps": 3750, "loss": 0.5124, "lr": 1.1516058604840891e-05, "epoch": 7.6, "percentage": 50.67, "elapsed_time": "0:51:05", "remaining_time": "0:49:44"}
+{"current_steps": 1910, "total_steps": 3750, "loss": 0.574, "lr": 1.1423985984577813e-05, "epoch": 7.64, "percentage": 50.93, "elapsed_time": "0:51:21", "remaining_time": "0:49:28"}
+{"current_steps": 1920, "total_steps": 3750, "loss": 0.5361, "lr": 1.1331789981472603e-05, "epoch": 7.68, "percentage": 51.2, "elapsed_time": "0:51:38", "remaining_time": "0:49:13"}
+{"current_steps": 1930, "total_steps": 3750, "loss": 0.5388, "lr": 1.1239478583950019e-05, "epoch": 7.72, "percentage": 51.47, "elapsed_time": "0:51:55", "remaining_time": "0:48:57"}
+{"current_steps": 1940, "total_steps": 3750, "loss": 0.536, "lr": 1.1147059790433296e-05, "epoch": 7.76, "percentage": 51.73, "elapsed_time": "0:52:11", "remaining_time": "0:48:41"}
+{"current_steps": 1950, "total_steps": 3750, "loss": 0.5165, "lr": 1.1054541608651121e-05, "epoch": 7.8, "percentage": 52.0, "elapsed_time": "0:52:27", "remaining_time": "0:48:25"}
+{"current_steps": 1960, "total_steps": 3750, "loss": 0.5369, "lr": 1.0961932054943778e-05, "epoch": 7.84, "percentage": 52.27, "elapsed_time": "0:52:43", "remaining_time": "0:48:09"}
+{"current_steps": 1970, "total_steps": 3750, "loss": 0.548, "lr": 1.0869239153568575e-05, "epoch": 7.88, "percentage": 52.53, "elapsed_time": "0:53:00", "remaining_time": "0:47:54"}
+{"current_steps": 1980, "total_steps": 3750, "loss": 0.5377, "lr": 1.0776470936004572e-05, "epoch": 7.92, "percentage": 52.8, "elapsed_time": "0:53:17", "remaining_time": "0:47:37"}
+{"current_steps": 1990, "total_steps": 3750, "loss": 0.5249, "lr": 1.0683635440256689e-05, "epoch": 7.96, "percentage": 53.07, "elapsed_time": "0:53:32", "remaining_time": "0:47:21"}
+{"current_steps": 2000, "total_steps": 3750, "loss": 0.5162, "lr": 1.059074071015923e-05, "epoch": 8.0, "percentage": 53.33, "elapsed_time": "0:53:49", "remaining_time": "0:47:05"}
+{"current_steps": 2010, "total_steps": 3750, "loss": 0.5067, "lr": 1.0497794794678923e-05, "epoch": 8.04, "percentage": 53.6, "elapsed_time": "0:54:05", "remaining_time": "0:46:49"}
+{"current_steps": 2020, "total_steps": 3750, "loss": 0.4901, "lr": 1.0404805747217525e-05, "epoch": 8.08, "percentage": 53.87, "elapsed_time": "0:54:21", "remaining_time": "0:46:33"}
+{"current_steps": 2030, "total_steps": 3750, "loss": 0.4834, "lr": 1.0311781624914e-05, "epoch": 8.12, "percentage": 54.13, "elapsed_time": "0:54:37", "remaining_time": "0:46:17"}
+{"current_steps": 2040, "total_steps": 3750, "loss": 0.4991, "lr": 1.0228036587536431e-05, "epoch": 8.16, "percentage": 54.4, "elapsed_time": "0:54:54", "remaining_time": "0:46:01"}
+{"current_steps": 2050, "total_steps": 3750, "loss": 0.5326, "lr": 1.013496803077246e-05, "epoch": 8.2, "percentage": 54.67, "elapsed_time": "0:55:11", "remaining_time": "0:45:45"}
+{"current_steps": 2060, "total_steps": 3750, "loss": 0.501, "lr": 1.0041887779554041e-05, "epoch": 8.24, "percentage": 54.93, "elapsed_time": "0:55:27", "remaining_time": "0:45:29"}
+{"current_steps": 2070, "total_steps": 3750, "loss": 0.5263, "lr": 9.948803898922586e-06, "epoch": 8.28, "percentage": 55.2, "elapsed_time": "0:55:43", "remaining_time": "0:45:13"}
+{"current_steps": 2080, "total_steps": 3750, "loss": 0.505, "lr": 9.85572445423399e-06, "epoch": 8.32, "percentage": 55.47, "elapsed_time": "0:55:59", "remaining_time": "0:44:56"}
+{"current_steps": 2090, "total_steps": 3750, "loss": 0.4962, "lr": 9.762657510459784e-06, "epoch": 8.36, "percentage": 55.73, "elapsed_time": "0:56:14", "remaining_time": "0:44:40"}
+{"current_steps": 2100, "total_steps": 3750, "loss": 0.5086, "lr": 9.669611131488346e-06, "epoch": 8.4, "percentage": 56.0, "elapsed_time": "0:56:30", "remaining_time": "0:44:23"}
+{"current_steps": 2110, "total_steps": 3750, "loss": 0.5105, "lr": 9.576593379426196e-06, "epoch": 8.44, "percentage": 56.27, "elapsed_time": "0:56:46", "remaining_time": "0:44:07"}
+{"current_steps": 2120, "total_steps": 3750, "loss": 0.5028, "lr": 9.483612313899436e-06, "epoch": 8.48, "percentage": 56.53, "elapsed_time": "0:57:03", "remaining_time": "0:43:51"}
+{"current_steps": 2130, "total_steps": 3750, "loss": 0.5273, "lr": 9.390675991355435e-06, "epoch": 8.52, "percentage": 56.8, "elapsed_time": "0:57:19", "remaining_time": "0:43:35"}
+{"current_steps": 2140, "total_steps": 3750, "loss": 0.4313, "lr": 9.297792464364748e-06, "epoch": 8.56, "percentage": 57.07, "elapsed_time": "0:57:34", "remaining_time": "0:43:18"}
+{"current_steps": 2150, "total_steps": 3750, "loss": 0.5045, "lr": 9.204969780923404e-06, "epoch": 8.6, "percentage": 57.33, "elapsed_time": "0:57:51", "remaining_time": "0:43:03"}
+{"current_steps": 2160, "total_steps": 3750, "loss": 0.4818, "lr": 9.112215983755573e-06, "epoch": 8.64, "percentage": 57.6, "elapsed_time": "0:58:07", "remaining_time": "0:42:47"}
+{"current_steps": 2170, "total_steps": 3750, "loss": 0.4779, "lr": 9.019539109616694e-06, "epoch": 8.68, "percentage": 57.87, "elapsed_time": "0:58:22", "remaining_time": "0:42:30"}
+{"current_steps": 2180, "total_steps": 3750, "loss": 0.4815, "lr": 8.926947188597133e-06, "epoch": 8.72, "percentage": 58.13, "elapsed_time": "0:58:38", "remaining_time": "0:42:14"}
+{"current_steps": 2190, "total_steps": 3750, "loss": 0.5259, "lr": 8.8344482434264e-06, "epoch": 8.76, "percentage": 58.4, "elapsed_time": "0:58:54", "remaining_time": "0:41:57"}
+{"current_steps": 2200, "total_steps": 3750, "loss": 0.5378, "lr": 8.742050288778e-06, "epoch": 8.8, "percentage": 58.67, "elapsed_time": "0:59:11", "remaining_time": "0:41:41"}
+{"current_steps": 2210, "total_steps": 3750, "loss": 0.5144, "lr": 8.649761330575009e-06, "epoch": 8.84, "percentage": 58.93, "elapsed_time": "0:59:28", "remaining_time": "0:41:26"}
+{"current_steps": 2220, "total_steps": 3750, "loss": 0.5383, "lr": 8.557589365296385e-06, "epoch": 8.88, "percentage": 59.2, "elapsed_time": "0:59:44", "remaining_time": "0:41:10"}
+{"current_steps": 2230, "total_steps": 3750, "loss": 0.4653, "lr": 8.4655423792841e-06, "epoch": 8.92, "percentage": 59.47, "elapsed_time": "1:00:01", "remaining_time": "0:40:54"}
+{"current_steps": 2240, "total_steps": 3750, "loss": 0.4868, "lr": 8.373628348051165e-06, "epoch": 8.96, "percentage": 59.73, "elapsed_time": "1:00:16", "remaining_time": "0:40:38"}
+{"current_steps": 2250, "total_steps": 3750, "loss": 0.5606, "lr": 8.281855235590574e-06, "epoch": 9.0, "percentage": 60.0, "elapsed_time": "1:00:33", "remaining_time": "0:40:22"}
+{"current_steps": 2260, "total_steps": 3750, "loss": 0.4717, "lr": 8.19023099368526e-06, "epoch": 9.04, "percentage": 60.27, "elapsed_time": "1:00:49", "remaining_time": "0:40:06"}
+{"current_steps": 2270, "total_steps": 3750, "loss": 0.4578, "lr": 8.098763561219101e-06, "epoch": 9.08, "percentage": 60.53, "elapsed_time": "1:01:05", "remaining_time": "0:39:49"}
+{"current_steps": 2280, "total_steps": 3750, "loss": 0.4553, "lr": 8.007460863489042e-06, "epoch": 9.12, "percentage": 60.8, "elapsed_time": "1:01:21", "remaining_time": "0:39:33"}
+{"current_steps": 2290, "total_steps": 3750, "loss": 0.4861, "lr": 7.91633081151841e-06, "epoch": 9.16, "percentage": 61.07, "elapsed_time": "1:01:37", "remaining_time": "0:39:17"}
+{"current_steps": 2300, "total_steps": 3750, "loss": 0.4518, "lr": 7.825381301371452e-06, "epoch": 9.2, "percentage": 61.33, "elapsed_time": "1:01:53", "remaining_time": "0:39:01"}
+{"current_steps": 2310, "total_steps": 3750, "loss": 0.4832, "lr": 7.734620213469166e-06, "epoch": 9.24, "percentage": 61.6, "elapsed_time": "1:02:09", "remaining_time": "0:38:44"}
+{"current_steps": 2320, "total_steps": 3750, "loss": 0.4969, "lr": 7.644055411906493e-06, "epoch": 9.28, "percentage": 61.87, "elapsed_time": "1:02:26", "remaining_time": "0:38:29"}
+{"current_steps": 2330, "total_steps": 3750, "loss": 0.4606, "lr": 7.553694743770928e-06, "epoch": 9.32, "percentage": 62.13, "elapsed_time": "1:02:42", "remaining_time": "0:38:12"}
+{"current_steps": 2340, "total_steps": 3750, "loss": 0.5225, "lr": 7.463546038462602e-06, "epoch": 9.36, "percentage": 62.4, "elapsed_time": "1:02:59", "remaining_time": "0:37:57"}
+{"current_steps": 2350, "total_steps": 3750, "loss": 0.529, "lr": 7.373617107015889e-06, "epoch": 9.4, "percentage": 62.67, "elapsed_time": "1:03:15", "remaining_time": "0:37:41"}
+{"current_steps": 2360, "total_steps": 3750, "loss": 0.5134, "lr": 7.283915741422611e-06, "epoch": 9.44, "percentage": 62.93, "elapsed_time": "1:03:32", "remaining_time": "0:37:25"}
+{"current_steps": 2370, "total_steps": 3750, "loss": 0.4509, "lr": 7.194449713956908e-06, "epoch": 9.48, "percentage": 63.2, "elapsed_time": "1:03:47", "remaining_time": "0:37:08"}
+{"current_steps": 2380, "total_steps": 3750, "loss": 0.5175, "lr": 7.105226776501772e-06, "epoch": 9.52, "percentage": 63.47, "elapsed_time": "1:04:03", "remaining_time": "0:36:52"}
+{"current_steps": 2390, "total_steps": 3750, "loss": 0.4742, "lr": 7.016254659877398e-06, "epoch": 9.56, "percentage": 63.73, "elapsed_time": "1:04:19", "remaining_time": "0:36:36"}
+{"current_steps": 2400, "total_steps": 3750, "loss": 0.4605, "lr": 6.927541073171333e-06, "epoch": 9.6, "percentage": 64.0, "elapsed_time": "1:04:35", "remaining_time": "0:36:19"}
+{"current_steps": 2410, "total_steps": 3750, "loss": 0.4987, "lr": 6.839093703070512e-06, "epoch": 9.64, "percentage": 64.27, "elapsed_time": "1:04:52", "remaining_time": "0:36:04"}
+{"current_steps": 2420, "total_steps": 3750, "loss": 0.4829, "lr": 6.750920213195238e-06, "epoch": 9.68, "percentage": 64.53, "elapsed_time": "1:05:08", "remaining_time": "0:35:47"}
+{"current_steps": 2430, "total_steps": 3750, "loss": 0.4842, "lr": 6.6630282434351535e-06, "epoch": 9.72, "percentage": 64.8, "elapsed_time": "1:05:23", "remaining_time": "0:35:31"}
+{"current_steps": 2440, "total_steps": 3750, "loss": 0.5198, "lr": 6.575425409287292e-06, "epoch": 9.76, "percentage": 65.07, "elapsed_time": "1:05:39", "remaining_time": "0:35:15"}
+{"current_steps": 2450, "total_steps": 3750, "loss": 0.5239, "lr": 6.488119301196201e-06, "epoch": 9.8, "percentage": 65.33, "elapsed_time": "1:05:55", "remaining_time": "0:34:59"}
+{"current_steps": 2460, "total_steps": 3750, "loss": 0.4636, "lr": 6.4011174838962706e-06, "epoch": 9.84, "percentage": 65.6, "elapsed_time": "1:06:11", "remaining_time": "0:34:42"}
+{"current_steps": 2470, "total_steps": 3750, "loss": 0.4693, "lr": 6.314427495756283e-06, "epoch": 9.88, "percentage": 65.87, "elapsed_time": "1:06:28", "remaining_time": "0:34:26"}
+{"current_steps": 2480, "total_steps": 3750, "loss": 0.475, "lr": 6.228056848126236e-06, "epoch": 9.92, "percentage": 66.13, "elapsed_time": "1:06:44", "remaining_time": "0:34:10"}
+{"current_steps": 2490, "total_steps": 3750, "loss": 0.4995, "lr": 6.142013024686509e-06, "epoch": 9.96, "percentage": 66.4, "elapsed_time": "1:07:00", "remaining_time": "0:33:54"}
+{"current_steps": 2500, "total_steps": 3750, "loss": 0.486, "lr": 6.056303480799449e-06, "epoch": 10.0, "percentage": 66.67, "elapsed_time": "1:07:17", "remaining_time": "0:33:38"}
+{"current_steps": 2510, "total_steps": 3750, "loss": 0.4537, "lr": 5.970935642863375e-06, "epoch": 10.04, "percentage": 66.93, "elapsed_time": "1:07:32", "remaining_time": "0:33:22"}
+{"current_steps": 2520, "total_steps": 3750, "loss": 0.3856, "lr": 5.885916907669114e-06, "epoch": 10.08, "percentage": 67.2, "elapsed_time": "1:07:48", "remaining_time": "0:33:05"}
+{"current_steps": 2530, "total_steps": 3750, "loss": 0.4705, "lr": 5.801254641759103e-06, "epoch": 10.12, "percentage": 67.47, "elapsed_time": "1:08:03", "remaining_time": "0:32:49"}
+{"current_steps": 2540, "total_steps": 3750, "loss": 0.5011, "lr": 5.716956180789098e-06, "epoch": 10.16, "percentage": 67.73, "elapsed_time": "1:08:20", "remaining_time": "0:32:33"}
+{"current_steps": 2550, "total_steps": 3750, "loss": 0.462, "lr": 5.6330288288925805e-06, "epoch": 10.2, "percentage": 68.0, "elapsed_time": "1:08:36", "remaining_time": "0:32:17"}
+{"current_steps": 2560, "total_steps": 3750, "loss": 0.5043, "lr": 5.549479858047875e-06, "epoch": 10.24, "percentage": 68.27, "elapsed_time": "1:08:53", "remaining_time": "0:32:01"}
+{"current_steps": 2570, "total_steps": 3750, "loss": 0.5244, "lr": 5.466316507448049e-06, "epoch": 10.28, "percentage": 68.53, "elapsed_time": "1:09:10", "remaining_time": "0:31:45"}
+{"current_steps": 2580, "total_steps": 3750, "loss": 0.4362, "lr": 5.3835459828736945e-06, "epoch": 10.32, "percentage": 68.8, "elapsed_time": "1:09:26", "remaining_time": "0:31:29"}
+{"current_steps": 2590, "total_steps": 3750, "loss": 0.4788, "lr": 5.30117545606854e-06, "epoch": 10.36, "percentage": 69.07, "elapsed_time": "1:09:42", "remaining_time": "0:31:13"}
+{"current_steps": 2600, "total_steps": 3750, "loss": 0.4265, "lr": 5.219212064118079e-06, "epoch": 10.4, "percentage": 69.33, "elapsed_time": "1:09:58", "remaining_time": "0:30:57"}
+{"current_steps": 2610, "total_steps": 3750, "loss": 0.5, "lr": 5.137662908831147e-06, "epoch": 10.44, "percentage": 69.6, "elapsed_time": "1:10:14", "remaining_time": "0:30:40"}
+{"current_steps": 2620, "total_steps": 3750, "loss": 0.4409, "lr": 5.056535056124592e-06, "epoch": 10.48, "percentage": 69.87, "elapsed_time": "1:10:30", "remaining_time": "0:30:24"}
+{"current_steps": 2630, "total_steps": 3750, "loss": 0.4594, "lr": 4.97583553541102e-06, "epoch": 10.52, "percentage": 70.13, "elapsed_time": "1:10:45", "remaining_time": "0:30:08"}
+{"current_steps": 2640, "total_steps": 3750, "loss": 0.4953, "lr": 4.895571338989754e-06, "epoch": 10.56, "percentage": 70.4, "elapsed_time": "1:11:01", "remaining_time": "0:29:51"}
+{"current_steps": 2650, "total_steps": 3750, "loss": 0.4795, "lr": 4.8157494214409475e-06, "epoch": 10.6, "percentage": 70.67, "elapsed_time": "1:11:18", "remaining_time": "0:29:35"}
+{"current_steps": 2660, "total_steps": 3750, "loss": 0.481, "lr": 4.736376699023023e-06, "epoch": 10.64, "percentage": 70.93, "elapsed_time": "1:11:35", "remaining_time": "0:29:19"}
+{"current_steps": 2670, "total_steps": 3750, "loss": 0.4713, "lr": 4.6574600490733794e-06, "epoch": 10.68, "percentage": 71.2, "elapsed_time": "1:11:51", "remaining_time": "0:29:03"}
+{"current_steps": 2680, "total_steps": 3750, "loss": 0.4501, "lr": 4.579006309412533e-06, "epoch": 10.72, "percentage": 71.47, "elapsed_time": "1:12:07", "remaining_time": "0:28:47"}
+{"current_steps": 2690, "total_steps": 3750, "loss": 0.4754, "lr": 4.501022277751602e-06, "epoch": 10.76, "percentage": 71.73, "elapsed_time": "1:12:23", "remaining_time": "0:28:31"}
+{"current_steps": 2700, "total_steps": 3750, "loss": 0.5056, "lr": 4.423514711103355e-06, "epoch": 10.8, "percentage": 72.0, "elapsed_time": "1:12:40", "remaining_time": "0:28:15"}
+{"current_steps": 2710, "total_steps": 3750, "loss": 0.4415, "lr": 4.346490325196704e-06, "epoch": 10.84, "percentage": 72.27, "elapsed_time": "1:12:56", "remaining_time": "0:27:59"}
+{"current_steps": 2720, "total_steps": 3750, "loss": 0.5117, "lr": 4.26995579389485e-06, "epoch": 10.88, "percentage": 72.53, "elapsed_time": "1:13:13", "remaining_time": "0:27:43"}
+{"current_steps": 2730, "total_steps": 3750, "loss": 0.475, "lr": 4.193917748616979e-06, "epoch": 10.92, "percentage": 72.8, "elapsed_time": "1:13:29", "remaining_time": "0:27:27"}
+{"current_steps": 2740, "total_steps": 3750, "loss": 0.4363, "lr": 4.118382777763711e-06, "epoch": 10.96, "percentage": 73.07, "elapsed_time": "1:13:45", "remaining_time": "0:27:11"}
+{"current_steps": 2750, "total_steps": 3750, "loss": 0.4665, "lr": 4.04335742614622e-06, "epoch": 11.0, "percentage": 73.33, "elapsed_time": "1:14:01", "remaining_time": "0:26:55"}
+{"current_steps": 2760, "total_steps": 3750, "loss": 0.4515, "lr": 3.968848194419163e-06, "epoch": 11.04, "percentage": 73.6, "elapsed_time": "1:14:17", "remaining_time": "0:26:38"}
+{"current_steps": 2770, "total_steps": 3750, "loss": 0.4285, "lr": 3.894861538517401e-06, "epoch": 11.08, "percentage": 73.87, "elapsed_time": "1:14:33", "remaining_time": "0:26:22"}
+{"current_steps": 2780, "total_steps": 3750, "loss": 0.4343, "lr": 3.821403869096658e-06, "epoch": 11.12, "percentage": 74.13, "elapsed_time": "1:14:49", "remaining_time": "0:26:06"}
+{"current_steps": 2790, "total_steps": 3750, "loss": 0.4766, "lr": 3.748481550978017e-06, "epoch": 11.16, "percentage": 74.4, "elapsed_time": "1:15:05", "remaining_time": "0:25:50"}
+{"current_steps": 2800, "total_steps": 3750, "loss": 0.4096, "lr": 3.6761009025964657e-06, "epoch": 11.2, "percentage": 74.67, "elapsed_time": "1:15:21", "remaining_time": "0:25:34"}
+{"current_steps": 2810, "total_steps": 3750, "loss": 0.4622, "lr": 3.604268195453421e-06, "epoch": 11.24, "percentage": 74.93, "elapsed_time": "1:15:37", "remaining_time": "0:25:17"}
+{"current_steps": 2820, "total_steps": 3750, "loss": 0.4437, "lr": 3.5329896535733133e-06, "epoch": 11.28, "percentage": 75.2, "elapsed_time": "1:15:53", "remaining_time": "0:25:01"}
+{"current_steps": 2830, "total_steps": 3750, "loss": 0.4871, "lr": 3.462271452964321e-06, "epoch": 11.32, "percentage": 75.47, "elapsed_time": "1:16:09", "remaining_time": "0:24:45"}
+{"current_steps": 2840, "total_steps": 3750, "loss": 0.4575, "lr": 3.3921197210832235e-06, "epoch": 11.36, "percentage": 75.73, "elapsed_time": "1:16:25", "remaining_time": "0:24:29"}
+{"current_steps": 2850, "total_steps": 3750, "loss": 0.4699, "lr": 3.3225405363045016e-06, "epoch": 11.4, "percentage": 76.0, "elapsed_time": "1:16:42", "remaining_time": "0:24:13"}
+{"current_steps": 2860, "total_steps": 3750, "loss": 0.4648, "lr": 3.2535399273936407e-06, "epoch": 11.44, "percentage": 76.27, "elapsed_time": "1:16:57", "remaining_time": "0:23:57"}
+{"current_steps": 2870, "total_steps": 3750, "loss": 0.3923, "lr": 3.1851238729848033e-06, "epoch": 11.48, "percentage": 76.53, "elapsed_time": "1:17:13", "remaining_time": "0:23:40"}
+{"current_steps": 2880, "total_steps": 3750, "loss": 0.4717, "lr": 3.11729830106276e-06, "epoch": 11.52, "percentage": 76.8, "elapsed_time": "1:17:31", "remaining_time": "0:23:25"}
+{"current_steps": 2890, "total_steps": 3750, "loss": 0.4556, "lr": 3.0500690884492836e-06, "epoch": 11.56, "percentage": 77.07, "elapsed_time": "1:17:47", "remaining_time": "0:23:09"}
+{"current_steps": 2900, "total_steps": 3750, "loss": 0.4785, "lr": 2.983442060293926e-06, "epoch": 11.6, "percentage": 77.33, "elapsed_time": "1:18:04", "remaining_time": "0:22:52"}
+{"current_steps": 2910, "total_steps": 3750, "loss": 0.463, "lr": 2.917422989569311e-06, "epoch": 11.64, "percentage": 77.6, "elapsed_time": "1:18:20", "remaining_time": "0:22:36"}
+{"current_steps": 2920, "total_steps": 3750, "loss": 0.4551, "lr": 2.852017596570901e-06, "epoch": 11.68, "percentage": 77.87, "elapsed_time": "1:18:36", "remaining_time": "0:22:20"}
+{"current_steps": 2930, "total_steps": 3750, "loss": 0.4501, "lr": 2.7872315484213954e-06, "epoch": 11.72, "percentage": 78.13, "elapsed_time": "1:18:52", "remaining_time": "0:22:04"}
+{"current_steps": 2940, "total_steps": 3750, "loss": 0.4338, "lr": 2.723070458579653e-06, "epoch": 11.76, "percentage": 78.4, "elapsed_time": "1:19:08", "remaining_time": "0:21:48"}
+{"current_steps": 2950, "total_steps": 3750, "loss": 0.4744, "lr": 2.6595398863543407e-06, "epoch": 11.8, "percentage": 78.67, "elapsed_time": "1:19:24", "remaining_time": "0:21:32"}
+{"current_steps": 2960, "total_steps": 3750, "loss": 0.4257, "lr": 2.596645336422219e-06, "epoch": 11.84, "percentage": 78.93, "elapsed_time": "1:19:40", "remaining_time": "0:21:15"}
+{"current_steps": 2970, "total_steps": 3750, "loss": 0.4676, "lr": 2.5343922583512026e-06, "epoch": 11.88, "percentage": 79.2, "elapsed_time": "1:19:56", "remaining_time": "0:20:59"}
+{"current_steps": 2980, "total_steps": 3750, "loss": 0.455, "lr": 2.472786046128156e-06, "epoch": 11.92, "percentage": 79.47, "elapsed_time": "1:20:13", "remaining_time": "0:20:43"}
+{"current_steps": 2990, "total_steps": 3750, "loss": 0.4646, "lr": 2.411832037691545e-06, "epoch": 11.96, "percentage": 79.73, "elapsed_time": "1:20:29", "remaining_time": "0:20:27"}
+{"current_steps": 3000, "total_steps": 3750, "loss": 0.4774, "lr": 2.3515355144689155e-06, "epoch": 12.0, "percentage": 80.0, "elapsed_time": "1:20:46", "remaining_time": "0:20:11"}
+{"current_steps": 3010, "total_steps": 3750, "loss": 0.4333, "lr": 2.2919017009192703e-06, "epoch": 12.04, "percentage": 80.27, "elapsed_time": "1:21:03", "remaining_time": "0:19:55"}
+{"current_steps": 3020, "total_steps": 3750, "loss": 0.456, "lr": 2.2329357640804118e-06, "epoch": 12.08, "percentage": 80.53, "elapsed_time": "1:21:19", "remaining_time": "0:19:39"}
+{"current_steps": 3030, "total_steps": 3750, "loss": 0.4054, "lr": 2.1746428131212126e-06, "epoch": 12.12, "percentage": 80.8, "elapsed_time": "1:21:34", "remaining_time": "0:19:23"}
+{"current_steps": 3040, "total_steps": 3750, "loss": 0.4875, "lr": 2.117027898898948e-06, "epoch": 12.16, "percentage": 81.07, "elapsed_time": "1:21:51", "remaining_time": "0:19:07"}
+{"current_steps": 3050, "total_steps": 3750, "loss": 0.4041, "lr": 2.0600960135216463e-06, "epoch": 12.2, "percentage": 81.33, "elapsed_time": "1:22:07", "remaining_time": "0:18:50"}
+{"current_steps": 3060, "total_steps": 3750, "loss": 0.5115, "lr": 2.003852089915548e-06, "epoch": 12.24, "percentage": 81.6, "elapsed_time": "1:22:23", "remaining_time": "0:18:34"}
+{"current_steps": 3070, "total_steps": 3750, "loss": 0.4459, "lr": 1.9483010013976766e-06, "epoch": 12.28, "percentage": 81.87, "elapsed_time": "1:22:40", "remaining_time": "0:18:18"}
+{"current_steps": 3080, "total_steps": 3750, "loss": 0.3677, "lr": 1.8934475612536019e-06, "epoch": 12.32, "percentage": 82.13, "elapsed_time": "1:22:55", "remaining_time": "0:18:02"}
+{"current_steps": 3090, "total_steps": 3750, "loss": 0.4353, "lr": 1.8392965223203707e-06, "epoch": 12.36, "percentage": 82.4, "elapsed_time": "1:23:11", "remaining_time": "0:17:46"}
+{"current_steps": 3100, "total_steps": 3750, "loss": 0.4578, "lr": 1.7858525765747047e-06, "epoch": 12.4, "percentage": 82.67, "elapsed_time": "1:23:27", "remaining_time": "0:17:29"}
+{"current_steps": 3110, "total_steps": 3750, "loss": 0.4057, "lr": 1.7331203547264452e-06, "epoch": 12.44, "percentage": 82.93, "elapsed_time": "1:23:43", "remaining_time": "0:17:13"}
+{"current_steps": 3120, "total_steps": 3750, "loss": 0.4532, "lr": 1.6811044258173425e-06, "epoch": 12.48, "percentage": 83.2, "elapsed_time": "1:24:00", "remaining_time": "0:16:57"}
+{"current_steps": 3130, "total_steps": 3750, "loss": 0.4551, "lr": 1.629809296825139e-06, "epoch": 12.52, "percentage": 83.47, "elapsed_time": "1:24:16", "remaining_time": "0:16:41"}
+{"current_steps": 3140, "total_steps": 3750, "loss": 0.4388, "lr": 1.579239412273078e-06, "epoch": 12.56, "percentage": 83.73, "elapsed_time": "1:24:32", "remaining_time": "0:16:25"}
+{"current_steps": 3150, "total_steps": 3750, "loss": 0.412, "lr": 1.5293991538447882e-06, "epoch": 12.6, "percentage": 84.0, "elapsed_time": "1:24:48", "remaining_time": "0:16:09"}
+{"current_steps": 3160, "total_steps": 3750, "loss": 0.4517, "lr": 1.4802928400046457e-06, "epoch": 12.64, "percentage": 84.27, "elapsed_time": "1:25:04", "remaining_time": "0:15:53"}
+{"current_steps": 3170, "total_steps": 3750, "loss": 0.4893, "lr": 1.4319247256235713e-06, "epoch": 12.68, "percentage": 84.53, "elapsed_time": "1:25:21", "remaining_time": "0:15:36"}
+{"current_steps": 3180, "total_steps": 3750, "loss": 0.4305, "lr": 1.3842990016103886e-06, "epoch": 12.72, "percentage": 84.8, "elapsed_time": "1:25:37", "remaining_time": "0:15:20"}
+{"current_steps": 3190, "total_steps": 3750, "loss": 0.3833, "lr": 1.3374197945486833e-06, "epoch": 12.76, "percentage": 85.07, "elapsed_time": "1:25:53", "remaining_time": "0:15:04"}
+{"current_steps": 3200, "total_steps": 3750, "loss": 0.4513, "lr": 1.2912911663392468e-06, "epoch": 12.8, "percentage": 85.33, "elapsed_time": "1:26:10", "remaining_time": "0:14:48"}
+{"current_steps": 3210, "total_steps": 3750, "loss": 0.4712, "lr": 1.245917113848144e-06, "epoch": 12.84, "percentage": 85.6, "elapsed_time": "1:26:26", "remaining_time": "0:14:32"}
+{"current_steps": 3220, "total_steps": 3750, "loss": 0.4789, "lr": 1.2013015685603813e-06, "epoch": 12.88, "percentage": 85.87, "elapsed_time": "1:26:43", "remaining_time": "0:14:16"}
+{"current_steps": 3230, "total_steps": 3750, "loss": 0.4128, "lr": 1.1574483962392768e-06, "epoch": 12.92, "percentage": 86.13, "elapsed_time": "1:26:59", "remaining_time": "0:14:00"}
+{"current_steps": 3240, "total_steps": 3750, "loss": 0.4949, "lr": 1.114361396591498e-06, "epoch": 12.96, "percentage": 86.4, "elapsed_time": "1:27:16", "remaining_time": "0:13:44"}
+{"current_steps": 3250, "total_steps": 3750, "loss": 0.4167, "lr": 1.0720443029378303e-06, "epoch": 13.0, "percentage": 86.67, "elapsed_time": "1:27:32", "remaining_time": "0:13:28"}
+{"current_steps": 3260, "total_steps": 3750, "loss": 0.4483, "lr": 1.0305007818897006e-06, "epoch": 13.04, "percentage": 86.93, "elapsed_time": "1:27:48", "remaining_time": "0:13:11"}
+{"current_steps": 3270, "total_steps": 3750, "loss": 0.454, "lr": 9.897344330314862e-07, "epoch": 13.08, "percentage": 87.2, "elapsed_time": "1:28:04", "remaining_time": "0:12:55"}
+{"current_steps": 3280, "total_steps": 3750, "loss": 0.4438, "lr": 9.497487886086132e-07, "epoch": 13.12, "percentage": 87.47, "elapsed_time": "1:28:20", "remaining_time": "0:12:39"}
+{"current_steps": 3290, "total_steps": 3750, "loss": 0.3904, "lr": 9.105473132215126e-07, "epoch": 13.16, "percentage": 87.73, "elapsed_time": "1:28:36", "remaining_time": "0:12:23"}
+{"current_steps": 3300, "total_steps": 3750, "loss": 0.4128, "lr": 8.721334035254203e-07, "epoch": 13.2, "percentage": 88.0, "elapsed_time": "1:28:52", "remaining_time": "0:12:07"}
+{"current_steps": 3310, "total_steps": 3750, "loss": 0.4479, "lr": 8.345103879360695e-07, "epoch": 13.24, "percentage": 88.27, "elapsed_time": "1:29:08", "remaining_time": "0:11:50"}
+{"current_steps": 3320, "total_steps": 3750, "loss": 0.3928, "lr": 7.976815263412963e-07, "epoch": 13.28, "percentage": 88.53, "elapsed_time": "1:29:24", "remaining_time": "0:11:34"}
+{"current_steps": 3330, "total_steps": 3750, "loss": 0.4163, "lr": 7.616500098185908e-07, "epoch": 13.32, "percentage": 88.8, "elapsed_time": "1:29:40", "remaining_time": "0:11:18"}
+{"current_steps": 3340, "total_steps": 3750, "loss": 0.4186, "lr": 7.264189603585892e-07, "epoch": 13.36, "percentage": 89.07, "elapsed_time": "1:29:56", "remaining_time": "0:11:02"}
+{"current_steps": 3350, "total_steps": 3750, "loss": 0.4416, "lr": 6.919914305945774e-07, "epoch": 13.4, "percentage": 89.33, "elapsed_time": "1:30:12", "remaining_time": "0:10:46"}
+{"current_steps": 3360, "total_steps": 3750, "loss": 0.3958, "lr": 6.58370403537989e-07, "epoch": 13.44, "percentage": 89.6, "elapsed_time": "1:30:28", "remaining_time": "0:10:30"}
+{"current_steps": 3370, "total_steps": 3750, "loss": 0.4424, "lr": 6.255587923199313e-07, "epoch": 13.48, "percentage": 89.87, "elapsed_time": "1:30:44", "remaining_time": "0:10:13"}
+{"current_steps": 3380, "total_steps": 3750, "loss": 0.4778, "lr": 5.935594399387856e-07, "epoch": 13.52, "percentage": 90.13, "elapsed_time": "1:31:01", "remaining_time": "0:09:57"}
+{"current_steps": 3390, "total_steps": 3750, "loss": 0.4045, "lr": 5.623751190138682e-07, "epoch": 13.56, "percentage": 90.4, "elapsed_time": "1:31:17", "remaining_time": "0:09:41"}
+{"current_steps": 3400, "total_steps": 3750, "loss": 0.4275, "lr": 5.320085315451862e-07, "epoch": 13.6, "percentage": 90.67, "elapsed_time": "1:31:33", "remaining_time": "0:09:25"}
+{"current_steps": 3410, "total_steps": 3750, "loss": 0.4346, "lr": 5.024623086793323e-07, "epoch": 13.64, "percentage": 90.93, "elapsed_time": "1:31:50", "remaining_time": "0:09:09"}
+{"current_steps": 3420, "total_steps": 3750, "loss": 0.4343, "lr": 4.737390104814954e-07, "epoch": 13.68, "percentage": 91.2, "elapsed_time": "1:32:06", "remaining_time": "0:08:53"}
+{"current_steps": 3430, "total_steps": 3750, "loss": 0.4355, "lr": 4.458411257136486e-07, "epoch": 13.72, "percentage": 91.47, "elapsed_time": "1:32:23", "remaining_time": "0:08:37"}
+{"current_steps": 3440, "total_steps": 3750, "loss": 0.4407, "lr": 4.1877107161890416e-07, "epoch": 13.76, "percentage": 91.73, "elapsed_time": "1:32:40", "remaining_time": "0:08:21"}
+{"current_steps": 3450, "total_steps": 3750, "loss": 0.4369, "lr": 3.9253119371206684e-07, "epoch": 13.8, "percentage": 92.0, "elapsed_time": "1:32:56", "remaining_time": "0:08:04"}
+{"current_steps": 3460, "total_steps": 3750, "loss": 0.4823, "lr": 3.671237655764104e-07, "epoch": 13.84, "percentage": 92.27, "elapsed_time": "1:33:13", "remaining_time": "0:07:48"}
+{"current_steps": 3470, "total_steps": 3750, "loss": 0.4566, "lr": 3.4255098866667114e-07, "epoch": 13.88, "percentage": 92.53, "elapsed_time": "1:33:29", "remaining_time": "0:07:32"}
+{"current_steps": 3480, "total_steps": 3750, "loss": 0.4824, "lr": 3.188149921183115e-07, "epoch": 13.92, "percentage": 92.8, "elapsed_time": "1:33:46", "remaining_time": "0:07:16"}
+{"current_steps": 3490, "total_steps": 3750, "loss": 0.3983, "lr": 2.959178325630296e-07, "epoch": 13.96, "percentage": 93.07, "elapsed_time": "1:34:02", "remaining_time": "0:07:00"}
+{"current_steps": 3500, "total_steps": 3750, "loss": 0.4541, "lr": 2.7386149395056463e-07, "epoch": 14.0, "percentage": 93.33, "elapsed_time": "1:34:18", "remaining_time": "0:06:44"}
+{"current_steps": 3510, "total_steps": 3750, "loss": 0.4667, "lr": 2.526478873767946e-07, "epoch": 14.04, "percentage": 93.6, "elapsed_time": "1:34:34", "remaining_time": "0:06:27"}
+{"current_steps": 3520, "total_steps": 3750, "loss": 0.441, "lr": 2.322788509181484e-07, "epoch": 14.08, "percentage": 93.87, "elapsed_time": "1:34:50", "remaining_time": "0:06:11"}
+{"current_steps": 3530, "total_steps": 3750, "loss": 0.4294, "lr": 2.1275614947233624e-07, "epoch": 14.12, "percentage": 94.13, "elapsed_time": "1:35:06", "remaining_time": "0:05:55"}
+{"current_steps": 3540, "total_steps": 3750, "loss": 0.4245, "lr": 1.9408147460544203e-07, "epoch": 14.16, "percentage": 94.4, "elapsed_time": "1:35:22", "remaining_time": "0:05:39"}
+{"current_steps": 3550, "total_steps": 3750, "loss": 0.3884, "lr": 1.7625644440534384e-07, "epoch": 14.2, "percentage": 94.67, "elapsed_time": "1:35:37", "remaining_time": "0:05:23"}
+{"current_steps": 3560, "total_steps": 3750, "loss": 0.4434, "lr": 1.5928260334151847e-07, "epoch": 14.24, "percentage": 94.93, "elapsed_time": "1:35:54", "remaining_time": "0:05:07"}
+{"current_steps": 3570, "total_steps": 3750, "loss": 0.4474, "lr": 1.4316142213121386e-07, "epoch": 14.28, "percentage": 95.2, "elapsed_time": "1:36:11", "remaining_time": "0:04:51"}
+{"current_steps": 3580, "total_steps": 3750, "loss": 0.3927, "lr": 1.2789429761202565e-07, "epoch": 14.32, "percentage": 95.47, "elapsed_time": "1:36:27", "remaining_time": "0:04:34"}
+{"current_steps": 3590, "total_steps": 3750, "loss": 0.3967, "lr": 1.134825526208605e-07, "epoch": 14.36, "percentage": 95.73, "elapsed_time": "1:36:43", "remaining_time": "0:04:18"}
+{"current_steps": 3600, "total_steps": 3750, "loss": 0.4807, "lr": 9.992743587931674e-08, "epoch": 14.4, "percentage": 96.0, "elapsed_time": "1:36:59", "remaining_time": "0:04:02"}
+{"current_steps": 3610, "total_steps": 3750, "loss": 0.4506, "lr": 8.723012188549318e-08, "epoch": 14.44, "percentage": 96.27, "elapsed_time": "1:37:15", "remaining_time": "0:03:46"}
+{"current_steps": 3620, "total_steps": 3750, "loss": 0.3934, "lr": 7.539171081221597e-08, "epoch": 14.48, "percentage": 96.53, "elapsed_time": "1:37:31", "remaining_time": "0:03:30"}
+{"current_steps": 3630, "total_steps": 3750, "loss": 0.4549, "lr": 6.44132284117216e-08, "epoch": 14.52, "percentage": 96.8, "elapsed_time": "1:37:48", "remaining_time": "0:03:13"}
+{"current_steps": 3640, "total_steps": 3750, "loss": 0.4208, "lr": 5.429562592677018e-08, "epoch": 14.56, "percentage": 97.07, "elapsed_time": "1:38:03", "remaining_time": "0:02:57"}
+{"current_steps": 3650, "total_steps": 3750, "loss": 0.4376, "lr": 4.503978000823028e-08, "epoch": 14.6, "percentage": 97.33, "elapsed_time": "1:38:20", "remaining_time": "0:02:41"}
+{"current_steps": 3660, "total_steps": 3750, "loss": 0.4562, "lr": 3.6646492639118567e-08, "epoch": 14.64, "percentage": 97.6, "elapsed_time": "1:38:36", "remaining_time": "0:02:25"}
+{"current_steps": 3670, "total_steps": 3750, "loss": 0.42, "lr": 2.911649106511316e-08, "epoch": 14.68, "percentage": 97.87, "elapsed_time": "1:38:54", "remaining_time": "0:02:09"}
+{"current_steps": 3680, "total_steps": 3750, "loss": 0.4489, "lr": 2.2450427731534052e-08, "epoch": 14.72, "percentage": 98.13, "elapsed_time": "1:39:10", "remaining_time": "0:01:53"}
+{"current_steps": 3690, "total_steps": 3750, "loss": 0.4532, "lr": 1.664888022682165e-08, "epoch": 14.76, "percentage": 98.4, "elapsed_time": "1:39:27", "remaining_time": "0:01:37"}
+{"current_steps": 3700, "total_steps": 3750, "loss": 0.4375, "lr": 1.1712351232480157e-08, "epoch": 14.8, "percentage": 98.67, "elapsed_time": "1:39:42", "remaining_time": "0:01:20"}
+{"current_steps": 3710, "total_steps": 3750, "loss": 0.4574, "lr": 7.641268479531283e-09, "epoch": 14.84, "percentage": 98.93, "elapsed_time": "1:39:58", "remaining_time": "0:01:04"}
+{"current_steps": 3720, "total_steps": 3750, "loss": 0.3538, "lr": 4.435984711446128e-09, "epoch": 14.88, "percentage": 99.2, "elapsed_time": "1:40:14", "remaining_time": "0:00:48"}
+{"current_steps": 3730, "total_steps": 3750, "loss": 0.3884, "lr": 2.0967776535851802e-09, "epoch": 14.92, "percentage": 99.47, "elapsed_time": "1:40:30", "remaining_time": "0:00:32"}
+{"current_steps": 3740, "total_steps": 3750, "loss": 0.4266, "lr": 6.238499891353389e-10, "epoch": 14.96, "percentage": 99.73, "elapsed_time": "1:40:46", "remaining_time": "0:00:16"}
+{"current_steps": 3750, "total_steps": 3750, "loss": 0.4743, "lr": 1.7329341542859922e-11, "epoch": 15.0, "percentage": 100.0, "elapsed_time": "1:41:02", "remaining_time": "0:00:00"}
+{"current_steps": 3750, "total_steps": 3750, "epoch": 15.0, "percentage": 100.0, "elapsed_time": "1:41:22", "remaining_time": "0:00:00"}

trainer_state.json

@@ -1,97 +1,2652 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
+  "epoch": 15.0,
   "eval_steps": 500,
-  "global_step": 100,
+  "global_step": 3750,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.1,
-      "grad_norm": 1.394713282585144,
-      "learning_rate": 2e-05,
-      "loss": 1.5103,
+      "epoch": 0.04,
+      "grad_norm": 3.0283896923065186,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 1.5494,
       "step": 10
     },
     {
-      "epoch": 0.2,
-      "grad_norm": 1.7702547311782837,
-      "learning_rate": 1.9396926207859085e-05,
-      "loss": 1.2272,
+      "epoch": 0.08,
+      "grad_norm": 2.980957508087158,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 1.5781,
       "step": 20
     },
     {
-      "epoch": 0.3,
-      "grad_norm": 1.7972954511642456,
-      "learning_rate": 1.766044443118978e-05,
-      "loss": 1.0918,
+      "epoch": 0.12,
+      "grad_norm": 1.3325070142745972,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.271,
       "step": 30
     },
     {
-      "epoch": 0.4,
-      "grad_norm": 1.4144309759140015,
-      "learning_rate": 1.5000000000000002e-05,
-      "loss": 1.0666,
+      "epoch": 0.16,
+      "grad_norm": 1.1161530017852783,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 1.1263,
       "step": 40
     },
     {
-      "epoch": 0.5,
-      "grad_norm": 1.6088738441467285,
-      "learning_rate": 1.1736481776669307e-05,
-      "loss": 0.9902,
+      "epoch": 0.2,
+      "grad_norm": 0.9847109317779541,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.4759,
       "step": 50
     },
     {
-      "epoch": 0.6,
-      "grad_norm": 1.517215609550476,
-      "learning_rate": 8.263518223330698e-06,
-      "loss": 0.8776,
+      "epoch": 0.24,
+      "grad_norm": 1.3176578283309937,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.1906,
       "step": 60
     },
     {
-      "epoch": 0.7,
-      "grad_norm": 1.463085651397705,
-      "learning_rate": 5.000000000000003e-06,
-      "loss": 0.8724,
+      "epoch": 0.28,
+      "grad_norm": 2.9614243507385254,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 1.3136,
       "step": 70
     },
     {
-      "epoch": 0.8,
-      "grad_norm": 1.6904792785644531,
-      "learning_rate": 2.339555568810221e-06,
-      "loss": 0.935,
+      "epoch": 0.32,
+      "grad_norm": 1.0635404586791992,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 1.159,
       "step": 80
     },
     {
-      "epoch": 0.9,
-      "grad_norm": 1.6216388940811157,
-      "learning_rate": 6.030737921409169e-07,
-      "loss": 0.9456,
+      "epoch": 0.36,
+      "grad_norm": 1.4873822927474976,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.0645,
       "step": 90
     },
     {
-      "epoch": 1.0,
-      "grad_norm": 1.4895377159118652,
-      "learning_rate": 0.0,
-      "loss": 0.8522,
+      "epoch": 0.4,
+      "grad_norm": 2.666663646697998,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.3174,
       "step": 100
     },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9259383678436279,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 1.0041,
+      "step": 110
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.422500729560852,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.1208,
+      "step": 120
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.513822317123413,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 0.9806,
+      "step": 130
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.26381254196167,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 0.9904,
+      "step": 140
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.899075984954834,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.9365,
+      "step": 150
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.7061178088188171,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 0.8843,
+      "step": 160
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.0236766338348389,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 0.9131,
+      "step": 170
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.3964245319366455,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.9142,
+      "step": 180
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7908815145492554,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 0.8973,
+      "step": 190
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.9264830946922302,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.8407,
+      "step": 200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5779850482940674,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.8228,
+      "step": 210
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6980341672897339,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 0.799,
+      "step": 220
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9194531440734863,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 0.7759,
+      "step": 230
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.0350003242492676,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.7466,
+      "step": 240
+    },
     {
       "epoch": 1.0,
-      "step": 100,
-      "total_flos": 2315448779538432.0,
-      "train_loss": 1.0368848323822022,
-      "train_runtime": 182.4958,
-      "train_samples_per_second": 1.096,
-      "train_steps_per_second": 0.548
+      "grad_norm": 3.474932909011841,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7317,
+      "step": 250
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.7944034337997437,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 0.7046,
+      "step": 260
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.945058584213257,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.6625,
+      "step": 270
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.820989966392517,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 0.7307,
+      "step": 280
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.9544613361358643,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 0.7996,
+      "step": 290
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9499707221984863,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.706,
+      "step": 300
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 5.53220796585083,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 0.6941,
+      "step": 310
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 2.7142622470855713,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 0.746,
+      "step": 320
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 4.010003089904785,
+      "learning_rate": 1.76e-05,
+      "loss": 0.7262,
+      "step": 330
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 2.3094098567962646,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 0.6552,
+      "step": 340
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 5.371938228607178,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.677,
+      "step": 350
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.662387728691101,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.6551,
+      "step": 360
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.2856159210205078,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 0.7076,
+      "step": 370
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2962098121643066,
+      "learning_rate": 1.999989169177959e-05,
+      "loss": 0.6818,
+      "step": 380
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.3778997659683228,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 0.6823,
+      "step": 390
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.4772292375564575,
+      "learning_rate": 1.999729241179462e-05,
+      "loss": 0.7704,
+      "step": 400
+    },
+    {
+      "epoch": 1.6400000000000001,
+      "grad_norm": 1.13938307762146,
+      "learning_rate": 1.999469335702714e-05,
+      "loss": 0.6668,
+      "step": 410
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 2.4890644550323486,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.6448,
+      "step": 420
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 2.2066543102264404,
+      "learning_rate": 1.998689754391257e-05,
+      "loss": 0.7159,
+      "step": 430
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.4606579542160034,
+      "learning_rate": 1.998170146104234e-05,
+      "loss": 0.6443,
+      "step": 440
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 5.692836284637451,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 0.7253,
+      "step": 450
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.7549424171447754,
+      "learning_rate": 1.9968715193738738e-05,
+      "loss": 0.6349,
+      "step": 460
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 2.5602545738220215,
+      "learning_rate": 1.9960926134514875e-05,
+      "loss": 0.6793,
+      "step": 470
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.393797755241394,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 0.6686,
+      "step": 480
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.6151896119117737,
+      "learning_rate": 1.9942759539322845e-05,
+      "loss": 0.6584,
+      "step": 490
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 2.0221006870269775,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.6771,
+      "step": 500
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.0078063011169434,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.6664,
+      "step": 510
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 2.788282871246338,
+      "learning_rate": 1.990905082010344e-05,
+      "loss": 0.6243,
+      "step": 520
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 2.064715623855591,
+      "learning_rate": 1.9896096046383456e-05,
+      "loss": 0.6253,
+      "step": 530
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 2.5293374061584473,
+      "learning_rate": 1.988228381446553e-05,
+      "loss": 0.6362,
+      "step": 540
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.461493730545044,
+      "learning_rate": 1.9867615321125796e-05,
+      "loss": 0.6517,
+      "step": 550
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.1433868408203125,
+      "learning_rate": 1.985209183733209e-05,
+      "loss": 0.6849,
+      "step": 560
+    },
+    {
+      "epoch": 2.2800000000000002,
+      "grad_norm": 1.6532901525497437,
+      "learning_rate": 1.983571470813386e-05,
+      "loss": 0.6298,
+      "step": 570
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 3.705383539199829,
+      "learning_rate": 1.9818485352545595e-05,
+      "loss": 0.6588,
+      "step": 580
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 2.4615492820739746,
+      "learning_rate": 1.980040526342388e-05,
+      "loss": 0.6154,
+      "step": 590
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.8189066052436829,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.6393,
+      "step": 600
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.1122651100158691,
+      "learning_rate": 1.9761699224434476e-05,
+      "loss": 0.6245,
+      "step": 610
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.4684017896652222,
+      "learning_rate": 1.9741076628294387e-05,
+      "loss": 0.6592,
+      "step": 620
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.9914065599441528,
+      "learning_rate": 1.9719610005785466e-05,
+      "loss": 0.6262,
+      "step": 630
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.7366482019424438,
+      "learning_rate": 1.969730121690698e-05,
+      "loss": 0.672,
+      "step": 640
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 3.544377326965332,
+      "learning_rate": 1.967415219462864e-05,
+      "loss": 0.6057,
+      "step": 650
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.9553754329681396,
+      "learning_rate": 1.9650164944723116e-05,
+      "loss": 0.6142,
+      "step": 660
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 2.1661672592163086,
+      "learning_rate": 1.9625341545592226e-05,
+      "loss": 0.6238,
+      "step": 670
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "grad_norm": 3.7167468070983887,
+      "learning_rate": 1.9599684148086876e-05,
+      "loss": 0.7166,
+      "step": 680
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 2.688824415206909,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.6769,
+      "step": 690
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 4.10930061340332,
+      "learning_rate": 1.954587632247732e-05,
+      "loss": 0.6199,
+      "step": 700
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.5201390981674194,
+      "learning_rate": 1.951773055661174e-05,
+      "loss": 0.6242,
+      "step": 710
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 3.6892731189727783,
+      "learning_rate": 1.9488760116444966e-05,
+      "loss": 0.6245,
+      "step": 720
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.8859150409698486,
+      "learning_rate": 1.9458967512152872e-05,
+      "loss": 0.628,
+      "step": 730
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.4320142269134521,
+      "learning_rate": 1.9428355325148632e-05,
+      "loss": 0.5806,
+      "step": 740
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.4816261529922485,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.5818,
+      "step": 750
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 2.1367580890655518,
+      "learning_rate": 1.9364682883494892e-05,
+      "loss": 0.6387,
+      "step": 760
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 2.2321407794952393,
+      "learning_rate": 1.9331628145814587e-05,
+      "loss": 0.6207,
+      "step": 770
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 3.410268783569336,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.5868,
+      "step": 780
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 3.323219060897827,
+      "learning_rate": 1.926309595682066e-05,
+      "loss": 0.5444,
+      "step": 790
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 2.397799015045166,
+      "learning_rate": 1.9227624443554425e-05,
+      "loss": 0.5891,
+      "step": 800
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 9.090506553649902,
+      "learning_rate": 1.9191353392552346e-05,
+      "loss": 0.5453,
+      "step": 810
+    },
+    {
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.3556101322174072,
+      "learning_rate": 1.9154285946559792e-05,
+      "loss": 0.6406,
+      "step": 820
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 1.056227684020996,
+      "learning_rate": 1.911642531732666e-05,
+      "loss": 0.5613,
+      "step": 830
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 1.2134612798690796,
+      "learning_rate": 1.907777478532909e-05,
+      "loss": 0.6439,
+      "step": 840
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 2.153582811355591,
+      "learning_rate": 1.9038337699485207e-05,
+      "loss": 0.6268,
+      "step": 850
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 1.4763509035110474,
+      "learning_rate": 1.8998117476864984e-05,
+      "loss": 0.6358,
+      "step": 860
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 2.120673656463623,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 0.5479,
+      "step": 870
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 3.2643983364105225,
+      "learning_rate": 1.8915341628552166e-05,
+      "loss": 0.5908,
+      "step": 880
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 1.7468228340148926,
+      "learning_rate": 1.8872793175064594e-05,
+      "loss": 0.6167,
+      "step": 890
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 1.6314669847488403,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 0.605,
+      "step": 900
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 1.2006853818893433,
+      "learning_rate": 1.8785393642396976e-05,
+      "loss": 0.6374,
+      "step": 910
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 5.40316915512085,
+      "learning_rate": 1.8740550136046195e-05,
+      "loss": 0.5919,
+      "step": 920
+    },
+    {
+      "epoch": 3.7199999999999998,
+      "grad_norm": 1.0310533046722412,
+      "learning_rate": 1.869494929505219e-05,
+      "loss": 0.596,
+      "step": 930
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 3.058582067489624,
+      "learning_rate": 1.8653264281300622e-05,
+      "loss": 0.5709,
+      "step": 940
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.8612210750579834,
+      "learning_rate": 1.8606235443821602e-05,
+      "loss": 0.6734,
+      "step": 950
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 1.0079221725463867,
+      "learning_rate": 1.8558460909544564e-05,
+      "loss": 0.5874,
+      "step": 960
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 1.8223471641540527,
+      "learning_rate": 1.850994481794692e-05,
+      "loss": 0.6199,
+      "step": 970
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 2.1497292518615723,
+      "learning_rate": 1.846069137275914e-05,
+      "loss": 0.5527,
+      "step": 980
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 1.5918537378311157,
+      "learning_rate": 1.8410704841600506e-05,
+      "loss": 0.5998,
+      "step": 990
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 4.21558952331543,
+      "learning_rate": 1.8359989555609355e-05,
+      "loss": 0.6357,
+      "step": 1000
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 2.512099504470825,
+      "learning_rate": 1.830854990906779e-05,
+      "loss": 0.6116,
+      "step": 1010
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 2.5760135650634766,
+      "learning_rate": 1.825639035902093e-05,
+      "loss": 0.54,
+      "step": 1020
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 1.5788276195526123,
+      "learning_rate": 1.8203515424890738e-05,
+      "loss": 0.6258,
+      "step": 1030
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 1.5123496055603027,
+      "learning_rate": 1.814992968808442e-05,
+      "loss": 0.5147,
+      "step": 1040
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 1.539919376373291,
+      "learning_rate": 1.809563779159746e-05,
+      "loss": 0.569,
+      "step": 1050
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 1.6823704242706299,
+      "learning_rate": 1.8040644439611348e-05,
+      "loss": 0.5588,
+      "step": 1060
+    },
+    {
+      "epoch": 4.28,
+      "grad_norm": 1.1113232374191284,
+      "learning_rate": 1.798495439708594e-05,
+      "loss": 0.5692,
+      "step": 1070
+    },
+    {
+      "epoch": 4.32,
+      "grad_norm": 1.6908786296844482,
+      "learning_rate": 1.792857248934663e-05,
+      "loss": 0.6102,
+      "step": 1080
+    },
+    {
+      "epoch": 4.36,
+      "grad_norm": 1.7746518850326538,
+      "learning_rate": 1.7871503601666233e-05,
+      "loss": 0.5706,
+      "step": 1090
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 1.2888718843460083,
+      "learning_rate": 1.7813752678841702e-05,
+      "loss": 0.5964,
+      "step": 1100
+    },
+    {
+      "epoch": 4.44,
+      "grad_norm": 2.7955427169799805,
+      "learning_rate": 1.7755324724765688e-05,
+      "loss": 0.6055,
+      "step": 1110
+    },
+    {
+      "epoch": 4.48,
+      "grad_norm": 1.4672576189041138,
+      "learning_rate": 1.7696224801992947e-05,
+      "loss": 0.5548,
+      "step": 1120
+    },
+    {
+      "epoch": 4.52,
+      "grad_norm": 2.2973852157592773,
+      "learning_rate": 1.7636458031301725e-05,
+      "loss": 0.5967,
+      "step": 1130
+    },
+    {
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.6734001636505127,
+      "learning_rate": 1.7576029591250036e-05,
+      "loss": 0.5567,
+      "step": 1140
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 2.127830743789673,
+      "learning_rate": 1.7514944717726962e-05,
+      "loss": 0.6065,
+      "step": 1150
+    },
+    {
+      "epoch": 4.64,
+      "grad_norm": 2.201108455657959,
+      "learning_rate": 1.7453208703499006e-05,
+      "loss": 0.566,
+      "step": 1160
+    },
+    {
+      "epoch": 4.68,
+      "grad_norm": 3.8374786376953125,
+      "learning_rate": 1.739082689775146e-05,
+      "loss": 0.55,
+      "step": 1170
+    },
+    {
+      "epoch": 4.72,
+      "grad_norm": 2.7282190322875977,
+      "learning_rate": 1.732780470562496e-05,
+      "loss": 0.5748,
+      "step": 1180
+    },
+    {
+      "epoch": 4.76,
+      "grad_norm": 1.8128880262374878,
+      "learning_rate": 1.7264147587747097e-05,
+      "loss": 0.6309,
+      "step": 1190
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 2.522096633911133,
+      "learning_rate": 1.7199861059759338e-05,
+      "loss": 0.5504,
+      "step": 1200
+    },
+    {
+      "epoch": 4.84,
+      "grad_norm": 1.186700463294983,
+      "learning_rate": 1.7134950691839063e-05,
+      "loss": 0.5741,
+      "step": 1210
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 4.312258720397949,
+      "learning_rate": 1.7069422108216973e-05,
+      "loss": 0.575,
+      "step": 1220
+    },
+    {
+      "epoch": 4.92,
+      "grad_norm": 3.402963399887085,
+      "learning_rate": 1.7003280986689733e-05,
+      "loss": 0.5842,
+      "step": 1230
+    },
+    {
+      "epoch": 4.96,
+      "grad_norm": 2.346266269683838,
+      "learning_rate": 1.693653305812805e-05,
+      "loss": 0.5877,
+      "step": 1240
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 1.304891586303711,
+      "learning_rate": 1.686918410598009e-05,
+      "loss": 0.6153,
+      "step": 1250
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 2.874284505844116,
+      "learning_rate": 1.6801239965770366e-05,
+      "loss": 0.5776,
+      "step": 1260
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 3.4333293437957764,
+      "learning_rate": 1.6732706524594138e-05,
+      "loss": 0.5099,
+      "step": 1270
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 1.4809489250183105,
+      "learning_rate": 1.6663589720607287e-05,
+      "loss": 0.5535,
+      "step": 1280
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 3.005042314529419,
+      "learning_rate": 1.659389554251181e-05,
+      "loss": 0.561,
+      "step": 1290
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 1.8546274900436401,
+      "learning_rate": 1.652363002903693e-05,
+      "loss": 0.555,
+      "step": 1300
+    },
+    {
+      "epoch": 5.24,
+      "grad_norm": 10.558731079101562,
+      "learning_rate": 1.6452799268415857e-05,
+      "loss": 0.5333,
+      "step": 1310
+    },
+    {
+      "epoch": 5.28,
+      "grad_norm": 2.2776925563812256,
+      "learning_rate": 1.6381409397858257e-05,
+      "loss": 0.5941,
+      "step": 1320
+    },
+    {
+      "epoch": 5.32,
+      "grad_norm": 2.6703994274139404,
+      "learning_rate": 1.6309466603018497e-05,
+      "loss": 0.5676,
+      "step": 1330
+    },
+    {
+      "epoch": 5.36,
+      "grad_norm": 3.8042726516723633,
+      "learning_rate": 1.6236977117459693e-05,
+      "loss": 0.5609,
+      "step": 1340
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 1.2701845169067383,
+      "learning_rate": 1.616394722211357e-05,
+      "loss": 0.5702,
+      "step": 1350
+    },
+    {
+      "epoch": 5.44,
+      "grad_norm": 1.392269253730774,
+      "learning_rate": 1.6090383244736256e-05,
+      "loss": 0.5388,
+      "step": 1360
+    },
+    {
+      "epoch": 5.48,
+      "grad_norm": 2.634445905685425,
+      "learning_rate": 1.6016291559360023e-05,
+      "loss": 0.573,
+      "step": 1370
+    },
+    {
+      "epoch": 5.52,
+      "grad_norm": 3.3800487518310547,
+      "learning_rate": 1.5941678585740976e-05,
+      "loss": 0.5522,
+      "step": 1380
+    },
+    {
+      "epoch": 5.5600000000000005,
+      "grad_norm": 2.935079336166382,
+      "learning_rate": 1.5866550788802815e-05,
+      "loss": 0.5615,
+      "step": 1390
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 1.9093431234359741,
+      "learning_rate": 1.579091467807668e-05,
+      "loss": 0.5537,
+      "step": 1400
+    },
+    {
+      "epoch": 5.64,
+      "grad_norm": 2.825533628463745,
+      "learning_rate": 1.5714776807137128e-05,
+      "loss": 0.6197,
+      "step": 1410
+    },
+    {
+      "epoch": 5.68,
+      "grad_norm": 3.172933578491211,
+      "learning_rate": 1.5638143773034268e-05,
+      "loss": 0.5415,
+      "step": 1420
+    },
+    {
+      "epoch": 5.72,
+      "grad_norm": 1.2932432889938354,
+      "learning_rate": 1.556102221572219e-05,
+      "loss": 0.5792,
+      "step": 1430
+    },
+    {
+      "epoch": 5.76,
+      "grad_norm": 2.1744470596313477,
+      "learning_rate": 1.5483418817483607e-05,
+      "loss": 0.52,
+      "step": 1440
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 2.4148924350738525,
+      "learning_rate": 1.540534030235087e-05,
+      "loss": 0.602,
+      "step": 1450
+    },
+    {
+      "epoch": 5.84,
+      "grad_norm": 2.427771806716919,
+      "learning_rate": 1.5326793435523374e-05,
+      "loss": 0.515,
+      "step": 1460
+    },
+    {
+      "epoch": 5.88,
+      "grad_norm": 1.566942811012268,
+      "learning_rate": 1.5247785022781343e-05,
+      "loss": 0.5795,
+      "step": 1470
+    },
+    {
+      "epoch": 5.92,
+      "grad_norm": 1.7555649280548096,
+      "learning_rate": 1.5168321909896171e-05,
+      "loss": 0.5819,
+      "step": 1480
+    },
+    {
+      "epoch": 5.96,
+      "grad_norm": 1.6367748975753784,
+      "learning_rate": 1.5088410982037251e-05,
+      "loss": 0.5244,
+      "step": 1490
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 1.9994490146636963,
+      "learning_rate": 1.50080591631754e-05,
+      "loss": 0.5823,
+      "step": 1500
+    },
+    {
+      "epoch": 6.04,
+      "grad_norm": 6.0024261474609375,
+      "learning_rate": 1.4927273415482916e-05,
+      "loss": 0.5641,
+      "step": 1510
+    },
+    {
+      "epoch": 6.08,
+      "grad_norm": 1.50034499168396,
+      "learning_rate": 1.484606073873035e-05,
+      "loss": 0.5325,
+      "step": 1520
+    },
+    {
+      "epoch": 6.12,
+      "grad_norm": 12.008216857910156,
+      "learning_rate": 1.4764428169679987e-05,
+      "loss": 0.5384,
+      "step": 1530
+    },
+    {
+      "epoch": 6.16,
+      "grad_norm": 1.6358847618103027,
+      "learning_rate": 1.4682382781476146e-05,
+      "loss": 0.5822,
+      "step": 1540
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 2.520883321762085,
+      "learning_rate": 1.4599931683032327e-05,
+      "loss": 0.5256,
+      "step": 1550
+    },
+    {
+      "epoch": 6.24,
+      "grad_norm": 1.136460304260254,
+      "learning_rate": 1.4517082018415231e-05,
+      "loss": 0.5589,
+      "step": 1560
+    },
+    {
+      "epoch": 6.28,
+      "grad_norm": 2.730435848236084,
+      "learning_rate": 1.4433840966225772e-05,
+      "loss": 0.4939,
+      "step": 1570
+    },
+    {
+      "epoch": 6.32,
+      "grad_norm": 0.9575507044792175,
+      "learning_rate": 1.4350215738977077e-05,
+      "loss": 0.5277,
+      "step": 1580
+    },
+    {
+      "epoch": 6.36,
+      "grad_norm": 1.9613964557647705,
+      "learning_rate": 1.4266213582469543e-05,
+      "loss": 0.5457,
+      "step": 1590
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 2.436429500579834,
+      "learning_rate": 1.4181841775163014e-05,
+      "loss": 0.547,
+      "step": 1600
+    },
+    {
+      "epoch": 6.44,
+      "grad_norm": 2.1270251274108887,
+      "learning_rate": 1.409710762754615e-05,
+      "loss": 0.5422,
+      "step": 1610
+    },
+    {
+      "epoch": 6.48,
+      "grad_norm": 2.4047000408172607,
+      "learning_rate": 1.4012018481502975e-05,
+      "loss": 0.5219,
+      "step": 1620
+    },
+    {
+      "epoch": 6.52,
+      "grad_norm": 1.1183472871780396,
+      "learning_rate": 1.3926581709676752e-05,
+      "loss": 0.5088,
+      "step": 1630
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.89410400390625,
+      "learning_rate": 1.3840804714831164e-05,
+      "loss": 0.5707,
+      "step": 1640
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 1.2478140592575073,
+      "learning_rate": 1.3754694929208891e-05,
+      "loss": 0.5893,
+      "step": 1650
+    },
+    {
+      "epoch": 6.64,
+      "grad_norm": 4.104971885681152,
+      "learning_rate": 1.3668259813887644e-05,
+      "loss": 0.5768,
+      "step": 1660
+    },
+    {
+      "epoch": 6.68,
+      "grad_norm": 1.4798212051391602,
+      "learning_rate": 1.3581506858133677e-05,
+      "loss": 0.5223,
+      "step": 1670
+    },
+    {
+      "epoch": 6.72,
+      "grad_norm": 3.716698408126831,
+      "learning_rate": 1.3494443578752893e-05,
+      "loss": 0.5208,
+      "step": 1680
+    },
+    {
+      "epoch": 6.76,
+      "grad_norm": 2.887411117553711,
+      "learning_rate": 1.340707751943952e-05,
+      "loss": 0.5415,
+      "step": 1690
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 3.6842939853668213,
+      "learning_rate": 1.3319416250122484e-05,
+      "loss": 0.5272,
+      "step": 1700
+    },
+    {
+      "epoch": 6.84,
+      "grad_norm": 4.174267292022705,
+      "learning_rate": 1.3231467366309523e-05,
+      "loss": 0.5341,
+      "step": 1710
+    },
+    {
+      "epoch": 6.88,
+      "grad_norm": 4.511580467224121,
+      "learning_rate": 1.3143238488429042e-05,
+      "loss": 0.573,
+      "step": 1720
+    },
+    {
+      "epoch": 6.92,
+      "grad_norm": 1.930474042892456,
+      "learning_rate": 1.3054737261169838e-05,
+      "loss": 0.5134,
+      "step": 1730
+    },
+    {
+      "epoch": 6.96,
+      "grad_norm": 1.2770944833755493,
+      "learning_rate": 1.2965971352818736e-05,
+      "loss": 0.4917,
+      "step": 1740
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 2.182473659515381,
+      "learning_rate": 1.287694845459613e-05,
+      "loss": 0.5725,
+      "step": 1750
+    },
+    {
+      "epoch": 7.04,
+      "grad_norm": 4.948633670806885,
+      "learning_rate": 1.2787676279989594e-05,
+      "loss": 0.5122,
+      "step": 1760
+    },
+    {
+      "epoch": 7.08,
+      "grad_norm": 2.5241198539733887,
+      "learning_rate": 1.2698162564085536e-05,
+      "loss": 0.4839,
+      "step": 1770
+    },
+    {
+      "epoch": 7.12,
+      "grad_norm": 1.8158693313598633,
+      "learning_rate": 1.2608415062898971e-05,
+      "loss": 0.4502,
+      "step": 1780
+    },
+    {
+      "epoch": 7.16,
+      "grad_norm": 3.540010929107666,
+      "learning_rate": 1.2518441552701493e-05,
+      "loss": 0.5585,
+      "step": 1790
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 1.9738972187042236,
+      "learning_rate": 1.2428249829347509e-05,
+      "loss": 0.4918,
+      "step": 1800
+    },
+    {
+      "epoch": 7.24,
+      "grad_norm": 1.0376015901565552,
+      "learning_rate": 1.2337847707598738e-05,
+      "loss": 0.4989,
+      "step": 1810
+    },
+    {
+      "epoch": 7.28,
+      "grad_norm": 2.5017688274383545,
+      "learning_rate": 1.2247243020447104e-05,
+      "loss": 0.4962,
+      "step": 1820
+    },
+    {
+      "epoch": 7.32,
+      "grad_norm": 1.2260102033615112,
+      "learning_rate": 1.2156443618436033e-05,
+      "loss": 0.5316,
+      "step": 1830
+    },
+    {
+      "epoch": 7.36,
+      "grad_norm": 2.4500176906585693,
+      "learning_rate": 1.2065457368980236e-05,
+      "loss": 0.4841,
+      "step": 1840
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 3.0221171379089355,
+      "learning_rate": 1.197429215568403e-05,
+      "loss": 0.5507,
+      "step": 1850
+    },
+    {
+      "epoch": 7.44,
+      "grad_norm": 5.26624059677124,
+      "learning_rate": 1.1882955877658252e-05,
+      "loss": 0.532,
+      "step": 1860
+    },
+    {
+      "epoch": 7.48,
+      "grad_norm": 2.411428213119507,
+      "learning_rate": 1.1791456448835825e-05,
+      "loss": 0.4802,
+      "step": 1870
+    },
+    {
+      "epoch": 7.52,
+      "grad_norm": 2.334620952606201,
+      "learning_rate": 1.169980179728606e-05,
+      "loss": 0.5331,
+      "step": 1880
+    },
+    {
+      "epoch": 7.5600000000000005,
+      "grad_norm": 2.4401047229766846,
+      "learning_rate": 1.1607999864527718e-05,
+      "loss": 0.4994,
+      "step": 1890
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 2.3867135047912598,
+      "learning_rate": 1.1516058604840891e-05,
+      "loss": 0.5124,
+      "step": 1900
+    },
+    {
+      "epoch": 7.64,
+      "grad_norm": 2.3309555053710938,
+      "learning_rate": 1.1423985984577813e-05,
+      "loss": 0.574,
+      "step": 1910
+    },
+    {
+      "epoch": 7.68,
+      "grad_norm": 1.1885383129119873,
+      "learning_rate": 1.1331789981472603e-05,
+      "loss": 0.5361,
+      "step": 1920
+    },
+    {
+      "epoch": 7.72,
+      "grad_norm": 1.6586416959762573,
+      "learning_rate": 1.1239478583950019e-05,
+      "loss": 0.5388,
+      "step": 1930
+    },
+    {
+      "epoch": 7.76,
+      "grad_norm": 1.3869335651397705,
+      "learning_rate": 1.1147059790433296e-05,
+      "loss": 0.536,
+      "step": 1940
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 1.5383076667785645,
+      "learning_rate": 1.1054541608651121e-05,
+      "loss": 0.5165,
+      "step": 1950
+    },
+    {
+      "epoch": 7.84,
+      "grad_norm": 1.1627497673034668,
+      "learning_rate": 1.0961932054943778e-05,
+      "loss": 0.5369,
+      "step": 1960
+    },
+    {
+      "epoch": 7.88,
+      "grad_norm": 1.4803476333618164,
+      "learning_rate": 1.0869239153568575e-05,
+      "loss": 0.548,
+      "step": 1970
+    },
+    {
+      "epoch": 7.92,
+      "grad_norm": 1.503915786743164,
+      "learning_rate": 1.0776470936004572e-05,
+      "loss": 0.5377,
+      "step": 1980
+    },
+    {
+      "epoch": 7.96,
+      "grad_norm": 1.9053574800491333,
+      "learning_rate": 1.0683635440256689e-05,
+      "loss": 0.5249,
+      "step": 1990
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 2.171719551086426,
+      "learning_rate": 1.059074071015923e-05,
+      "loss": 0.5162,
+      "step": 2000
+    },
+    {
+      "epoch": 8.04,
+      "grad_norm": 3.7397103309631348,
+      "learning_rate": 1.0497794794678923e-05,
+      "loss": 0.5067,
+      "step": 2010
+    },
+    {
+      "epoch": 8.08,
+      "grad_norm": 3.237569570541382,
+      "learning_rate": 1.0404805747217525e-05,
+      "loss": 0.4901,
+      "step": 2020
+    },
+    {
+      "epoch": 8.12,
+      "grad_norm": 2.6131529808044434,
+      "learning_rate": 1.0311781624914e-05,
+      "loss": 0.4834,
+      "step": 2030
+    },
+    {
+      "epoch": 8.16,
+      "grad_norm": 2.543020009994507,
+      "learning_rate": 1.0228036587536431e-05,
+      "loss": 0.4991,
+      "step": 2040
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 2.420510768890381,
+      "learning_rate": 1.013496803077246e-05,
+      "loss": 0.5326,
+      "step": 2050
+    },
+    {
+      "epoch": 8.24,
+      "grad_norm": 1.7979626655578613,
+      "learning_rate": 1.0041887779554041e-05,
+      "loss": 0.501,
+      "step": 2060
+    },
+    {
+      "epoch": 8.28,
+      "grad_norm": 3.0351650714874268,
+      "learning_rate": 9.948803898922586e-06,
+      "loss": 0.5263,
+      "step": 2070
+    },
+    {
+      "epoch": 8.32,
+      "grad_norm": 2.1602799892425537,
+      "learning_rate": 9.85572445423399e-06,
+      "loss": 0.505,
+      "step": 2080
+    },
+    {
+      "epoch": 8.36,
+      "grad_norm": 2.298388957977295,
+      "learning_rate": 9.762657510459784e-06,
+      "loss": 0.4962,
+      "step": 2090
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 1.9878581762313843,
+      "learning_rate": 9.669611131488346e-06,
+      "loss": 0.5086,
+      "step": 2100
+    },
+    {
+      "epoch": 8.44,
+      "grad_norm": 3.1122074127197266,
+      "learning_rate": 9.576593379426196e-06,
+      "loss": 0.5105,
+      "step": 2110
+    },
+    {
+      "epoch": 8.48,
+      "grad_norm": 1.8491990566253662,
+      "learning_rate": 9.483612313899436e-06,
+      "loss": 0.5028,
+      "step": 2120
+    },
+    {
+      "epoch": 8.52,
+      "grad_norm": 2.2476413249969482,
+      "learning_rate": 9.390675991355435e-06,
+      "loss": 0.5273,
+      "step": 2130
+    },
+    {
+      "epoch": 8.56,
+      "grad_norm": 3.6653342247009277,
+      "learning_rate": 9.297792464364748e-06,
+      "loss": 0.4313,
+      "step": 2140
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.8962536454200745,
+      "learning_rate": 9.204969780923404e-06,
+      "loss": 0.5045,
+      "step": 2150
+    },
+    {
+      "epoch": 8.64,
+      "grad_norm": 6.7541823387146,
+      "learning_rate": 9.112215983755573e-06,
+      "loss": 0.4818,
+      "step": 2160
+    },
+    {
+      "epoch": 8.68,
+      "grad_norm": 3.15523362159729,
+      "learning_rate": 9.019539109616694e-06,
+      "loss": 0.4779,
+      "step": 2170
+    },
+    {
+      "epoch": 8.72,
+      "grad_norm": 1.048300862312317,
+      "learning_rate": 8.926947188597133e-06,
+      "loss": 0.4815,
+      "step": 2180
+    },
+    {
+      "epoch": 8.76,
+      "grad_norm": 4.415710926055908,
+      "learning_rate": 8.8344482434264e-06,
+      "loss": 0.5259,
+      "step": 2190
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 4.474966049194336,
+      "learning_rate": 8.742050288778e-06,
+      "loss": 0.5378,
+      "step": 2200
+    },
+    {
+      "epoch": 8.84,
+      "grad_norm": 3.487746477127075,
+      "learning_rate": 8.649761330575009e-06,
+      "loss": 0.5144,
+      "step": 2210
+    },
+    {
+      "epoch": 8.88,
+      "grad_norm": 1.44117271900177,
+      "learning_rate": 8.557589365296385e-06,
+      "loss": 0.5383,
+      "step": 2220
+    },
+    {
+      "epoch": 8.92,
+      "grad_norm": 2.9286913871765137,
+      "learning_rate": 8.4655423792841e-06,
+      "loss": 0.4653,
+      "step": 2230
+    },
+    {
+      "epoch": 8.96,
+      "grad_norm": 1.4172818660736084,
+      "learning_rate": 8.373628348051165e-06,
+      "loss": 0.4868,
+      "step": 2240
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 1.9049030542373657,
+      "learning_rate": 8.281855235590574e-06,
+      "loss": 0.5606,
+      "step": 2250
+    },
+    {
+      "epoch": 9.04,
+      "grad_norm": 3.6202874183654785,
+      "learning_rate": 8.19023099368526e-06,
+      "loss": 0.4717,
+      "step": 2260
+    },
+    {
+      "epoch": 9.08,
+      "grad_norm": 1.4381736516952515,
+      "learning_rate": 8.098763561219101e-06,
+      "loss": 0.4578,
+      "step": 2270
+    },
+    {
+      "epoch": 9.12,
+      "grad_norm": 3.8551642894744873,
+      "learning_rate": 8.007460863489042e-06,
+      "loss": 0.4553,
+      "step": 2280
+    },
+    {
+      "epoch": 9.16,
+      "grad_norm": 2.2333943843841553,
+      "learning_rate": 7.91633081151841e-06,
+      "loss": 0.4861,
+      "step": 2290
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 3.517455816268921,
+      "learning_rate": 7.825381301371452e-06,
+      "loss": 0.4518,
+      "step": 2300
+    },
+    {
+      "epoch": 9.24,
+      "grad_norm": 1.2912664413452148,
+      "learning_rate": 7.734620213469166e-06,
+      "loss": 0.4832,
+      "step": 2310
+    },
+    {
+      "epoch": 9.28,
+      "grad_norm": 3.6948964595794678,
+      "learning_rate": 7.644055411906493e-06,
+      "loss": 0.4969,
+      "step": 2320
+    },
+    {
+      "epoch": 9.32,
+      "grad_norm": 1.5953376293182373,
+      "learning_rate": 7.553694743770928e-06,
+      "loss": 0.4606,
+      "step": 2330
+    },
+    {
+      "epoch": 9.36,
+      "grad_norm": 2.7939870357513428,
+      "learning_rate": 7.463546038462602e-06,
+      "loss": 0.5225,
+      "step": 2340
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 1.0297088623046875,
+      "learning_rate": 7.373617107015889e-06,
+      "loss": 0.529,
+      "step": 2350
+    },
+    {
+      "epoch": 9.44,
+      "grad_norm": 2.87479305267334,
+      "learning_rate": 7.283915741422611e-06,
+      "loss": 0.5134,
+      "step": 2360
+    },
+    {
+      "epoch": 9.48,
+      "grad_norm": 3.1623082160949707,
+      "learning_rate": 7.194449713956908e-06,
+      "loss": 0.4509,
+      "step": 2370
+    },
+    {
+      "epoch": 9.52,
+      "grad_norm": 1.8917375802993774,
+      "learning_rate": 7.105226776501772e-06,
+      "loss": 0.5175,
+      "step": 2380
+    },
+    {
+      "epoch": 9.56,
+      "grad_norm": 1.6095237731933594,
+      "learning_rate": 7.016254659877398e-06,
+      "loss": 0.4742,
+      "step": 2390
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 3.2498207092285156,
+      "learning_rate": 6.927541073171333e-06,
+      "loss": 0.4605,
+      "step": 2400
+    },
+    {
+      "epoch": 9.64,
+      "grad_norm": 1.7395751476287842,
+      "learning_rate": 6.839093703070512e-06,
+      "loss": 0.4987,
+      "step": 2410
+    },
+    {
+      "epoch": 9.68,
+      "grad_norm": 2.4571480751037598,
+      "learning_rate": 6.750920213195238e-06,
+      "loss": 0.4829,
+      "step": 2420
+    },
+    {
+      "epoch": 9.72,
+      "grad_norm": 4.019631385803223,
+      "learning_rate": 6.6630282434351535e-06,
+      "loss": 0.4842,
+      "step": 2430
+    },
+    {
+      "epoch": 9.76,
+      "grad_norm": 2.756540298461914,
+      "learning_rate": 6.575425409287292e-06,
+      "loss": 0.5198,
+      "step": 2440
+    },
+    {
+      "epoch": 9.8,
+      "grad_norm": 2.0040042400360107,
+      "learning_rate": 6.488119301196201e-06,
+      "loss": 0.5239,
+      "step": 2450
+    },
+    {
+      "epoch": 9.84,
+      "grad_norm": 7.419244766235352,
+      "learning_rate": 6.4011174838962706e-06,
+      "loss": 0.4636,
+      "step": 2460
+    },
+    {
+      "epoch": 9.88,
+      "grad_norm": 2.879230260848999,
+      "learning_rate": 6.314427495756283e-06,
+      "loss": 0.4693,
+      "step": 2470
+    },
+    {
+      "epoch": 9.92,
+      "grad_norm": 2.1217892169952393,
+      "learning_rate": 6.228056848126236e-06,
+      "loss": 0.475,
+      "step": 2480
+    },
+    {
+      "epoch": 9.96,
+      "grad_norm": 2.1474809646606445,
+      "learning_rate": 6.142013024686509e-06,
+      "loss": 0.4995,
+      "step": 2490
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 4.3783721923828125,
+      "learning_rate": 6.056303480799449e-06,
+      "loss": 0.486,
+      "step": 2500
+    },
+    {
+      "epoch": 10.04,
+      "grad_norm": 1.397594928741455,
+      "learning_rate": 5.970935642863375e-06,
+      "loss": 0.4537,
+      "step": 2510
+    },
+    {
+      "epoch": 10.08,
+      "grad_norm": 3.2472903728485107,
+      "learning_rate": 5.885916907669114e-06,
+      "loss": 0.3856,
+      "step": 2520
+    },
+    {
+      "epoch": 10.12,
+      "grad_norm": 3.1067910194396973,
+      "learning_rate": 5.801254641759103e-06,
+      "loss": 0.4705,
+      "step": 2530
+    },
+    {
+      "epoch": 10.16,
+      "grad_norm": 2.5792055130004883,
+      "learning_rate": 5.716956180789098e-06,
+      "loss": 0.5011,
+      "step": 2540
+    },
+    {
+      "epoch": 10.2,
+      "grad_norm": 3.1249446868896484,
+      "learning_rate": 5.6330288288925805e-06,
+      "loss": 0.462,
+      "step": 2550
+    },
+    {
+      "epoch": 10.24,
+      "grad_norm": 4.708195209503174,
+      "learning_rate": 5.549479858047875e-06,
+      "loss": 0.5043,
+      "step": 2560
+    },
+    {
+      "epoch": 10.28,
+      "grad_norm": 1.4850634336471558,
+      "learning_rate": 5.466316507448049e-06,
+      "loss": 0.5244,
+      "step": 2570
+    },
+    {
+      "epoch": 10.32,
+      "grad_norm": 1.0298686027526855,
+      "learning_rate": 5.3835459828736945e-06,
+      "loss": 0.4362,
+      "step": 2580
+    },
+    {
+      "epoch": 10.36,
+      "grad_norm": 2.251105546951294,
+      "learning_rate": 5.30117545606854e-06,
+      "loss": 0.4788,
+      "step": 2590
+    },
+    {
+      "epoch": 10.4,
+      "grad_norm": 4.622702121734619,
+      "learning_rate": 5.219212064118079e-06,
+      "loss": 0.4265,
+      "step": 2600
+    },
+    {
+      "epoch": 10.44,
+      "grad_norm": 3.4977996349334717,
+      "learning_rate": 5.137662908831147e-06,
+      "loss": 0.5,
+      "step": 2610
+    },
+    {
+      "epoch": 10.48,
+      "grad_norm": 5.369349002838135,
+      "learning_rate": 5.056535056124592e-06,
+      "loss": 0.4409,
+      "step": 2620
+    },
+    {
+      "epoch": 10.52,
+      "grad_norm": 2.318140983581543,
+      "learning_rate": 4.97583553541102e-06,
+      "loss": 0.4594,
+      "step": 2630
+    },
+    {
+      "epoch": 10.56,
+      "grad_norm": 2.4116406440734863,
+      "learning_rate": 4.895571338989754e-06,
+      "loss": 0.4953,
+      "step": 2640
+    },
+    {
+      "epoch": 10.6,
+      "grad_norm": 3.0506629943847656,
+      "learning_rate": 4.8157494214409475e-06,
+      "loss": 0.4795,
+      "step": 2650
+    },
+    {
+      "epoch": 10.64,
+      "grad_norm": 1.8125630617141724,
+      "learning_rate": 4.736376699023023e-06,
+      "loss": 0.481,
+      "step": 2660
+    },
+    {
+      "epoch": 10.68,
+      "grad_norm": 1.9686360359191895,
+      "learning_rate": 4.6574600490733794e-06,
+      "loss": 0.4713,
+      "step": 2670
+    },
+    {
+      "epoch": 10.72,
+      "grad_norm": 2.8264060020446777,
+      "learning_rate": 4.579006309412533e-06,
+      "loss": 0.4501,
+      "step": 2680
+    },
+    {
+      "epoch": 10.76,
+      "grad_norm": 1.9835346937179565,
+      "learning_rate": 4.501022277751602e-06,
+      "loss": 0.4754,
+      "step": 2690
+    },
+    {
+      "epoch": 10.8,
+      "grad_norm": 4.487490653991699,
+      "learning_rate": 4.423514711103355e-06,
+      "loss": 0.5056,
+      "step": 2700
+    },
+    {
+      "epoch": 10.84,
+      "grad_norm": 3.1984522342681885,
+      "learning_rate": 4.346490325196704e-06,
+      "loss": 0.4415,
+      "step": 2710
+    },
+    {
+      "epoch": 10.88,
+      "grad_norm": 2.0367348194122314,
+      "learning_rate": 4.26995579389485e-06,
+      "loss": 0.5117,
+      "step": 2720
+    },
+    {
+      "epoch": 10.92,
+      "grad_norm": 1.78911292552948,
+      "learning_rate": 4.193917748616979e-06,
+      "loss": 0.475,
+      "step": 2730
+    },
+    {
+      "epoch": 10.96,
+      "grad_norm": 2.0589475631713867,
+      "learning_rate": 4.118382777763711e-06,
+      "loss": 0.4363,
+      "step": 2740
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 3.54664945602417,
+      "learning_rate": 4.04335742614622e-06,
+      "loss": 0.4665,
+      "step": 2750
+    },
+    {
+      "epoch": 11.04,
+      "grad_norm": 1.714920997619629,
+      "learning_rate": 3.968848194419163e-06,
+      "loss": 0.4515,
+      "step": 2760
+    },
+    {
+      "epoch": 11.08,
+      "grad_norm": 11.161652565002441,
+      "learning_rate": 3.894861538517401e-06,
+      "loss": 0.4285,
+      "step": 2770
+    },
+    {
+      "epoch": 11.12,
+      "grad_norm": 2.627831220626831,
+      "learning_rate": 3.821403869096658e-06,
+      "loss": 0.4343,
+      "step": 2780
+    },
+    {
+      "epoch": 11.16,
+      "grad_norm": 2.6865172386169434,
+      "learning_rate": 3.748481550978017e-06,
+      "loss": 0.4766,
+      "step": 2790
+    },
+    {
+      "epoch": 11.2,
+      "grad_norm": 4.996657848358154,
+      "learning_rate": 3.6761009025964657e-06,
+      "loss": 0.4096,
+      "step": 2800
+    },
+    {
+      "epoch": 11.24,
+      "grad_norm": 1.6282066106796265,
+      "learning_rate": 3.604268195453421e-06,
+      "loss": 0.4622,
+      "step": 2810
+    },
+    {
+      "epoch": 11.28,
+      "grad_norm": 1.6030402183532715,
+      "learning_rate": 3.5329896535733133e-06,
+      "loss": 0.4437,
+      "step": 2820
+    },
+    {
+      "epoch": 11.32,
+      "grad_norm": 2.8916399478912354,
+      "learning_rate": 3.462271452964321e-06,
+      "loss": 0.4871,
+      "step": 2830
+    },
+    {
+      "epoch": 11.36,
+      "grad_norm": 2.375190019607544,
+      "learning_rate": 3.3921197210832235e-06,
+      "loss": 0.4575,
+      "step": 2840
+    },
+    {
+      "epoch": 11.4,
+      "grad_norm": 4.021700382232666,
+      "learning_rate": 3.3225405363045016e-06,
+      "loss": 0.4699,
+      "step": 2850
+    },
+    {
+      "epoch": 11.44,
+      "grad_norm": 1.7844669818878174,
+      "learning_rate": 3.2535399273936407e-06,
+      "loss": 0.4648,
+      "step": 2860
+    },
+    {
+      "epoch": 11.48,
+      "grad_norm": 2.744528293609619,
+      "learning_rate": 3.1851238729848033e-06,
+      "loss": 0.3923,
+      "step": 2870
+    },
+    {
+      "epoch": 11.52,
+      "grad_norm": 1.6203703880310059,
+      "learning_rate": 3.11729830106276e-06,
+      "loss": 0.4717,
+      "step": 2880
+    },
+    {
+      "epoch": 11.56,
+      "grad_norm": 1.696370244026184,
+      "learning_rate": 3.0500690884492836e-06,
+      "loss": 0.4556,
+      "step": 2890
+    },
+    {
+      "epoch": 11.6,
+      "grad_norm": 4.04744291305542,
+      "learning_rate": 2.983442060293926e-06,
+      "loss": 0.4785,
+      "step": 2900
+    },
+    {
+      "epoch": 11.64,
+      "grad_norm": 2.629739284515381,
+      "learning_rate": 2.917422989569311e-06,
+      "loss": 0.463,
+      "step": 2910
+    },
+    {
+      "epoch": 11.68,
+      "grad_norm": 2.43945050239563,
+      "learning_rate": 2.852017596570901e-06,
+      "loss": 0.4551,
+      "step": 2920
+    },
+    {
+      "epoch": 11.72,
+      "grad_norm": 6.5788116455078125,
+      "learning_rate": 2.7872315484213954e-06,
+      "loss": 0.4501,
+      "step": 2930
+    },
+    {
+      "epoch": 11.76,
+      "grad_norm": 2.3283305168151855,
+      "learning_rate": 2.723070458579653e-06,
+      "loss": 0.4338,
+      "step": 2940
+    },
+    {
+      "epoch": 11.8,
+      "grad_norm": 4.168436527252197,
+      "learning_rate": 2.6595398863543407e-06,
+      "loss": 0.4744,
+      "step": 2950
+    },
+    {
+      "epoch": 11.84,
+      "grad_norm": 3.3579213619232178,
+      "learning_rate": 2.596645336422219e-06,
+      "loss": 0.4257,
+      "step": 2960
+    },
+    {
+      "epoch": 11.88,
+      "grad_norm": 4.208755970001221,
+      "learning_rate": 2.5343922583512026e-06,
+      "loss": 0.4676,
+      "step": 2970
+    },
+    {
+      "epoch": 11.92,
+      "grad_norm": 2.752279281616211,
+      "learning_rate": 2.472786046128156e-06,
+      "loss": 0.455,
+      "step": 2980
+    },
+    {
+      "epoch": 11.96,
+      "grad_norm": 3.5390079021453857,
+      "learning_rate": 2.411832037691545e-06,
+      "loss": 0.4646,
+      "step": 2990
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 2.805065870285034,
+      "learning_rate": 2.3515355144689155e-06,
+      "loss": 0.4774,
+      "step": 3000
+    },
+    {
+      "epoch": 12.04,
+      "grad_norm": 2.3203487396240234,
+      "learning_rate": 2.2919017009192703e-06,
+      "loss": 0.4333,
+      "step": 3010
+    },
+    {
+      "epoch": 12.08,
+      "grad_norm": 2.62695050239563,
+      "learning_rate": 2.2329357640804118e-06,
+      "loss": 0.456,
+      "step": 3020
+    },
+    {
+      "epoch": 12.12,
+      "grad_norm": 1.8128643035888672,
+      "learning_rate": 2.1746428131212126e-06,
+      "loss": 0.4054,
+      "step": 3030
+    },
+    {
+      "epoch": 12.16,
+      "grad_norm": 1.5641071796417236,
+      "learning_rate": 2.117027898898948e-06,
+      "loss": 0.4875,
+      "step": 3040
+    },
+    {
+      "epoch": 12.2,
+      "grad_norm": 2.1929521560668945,
+      "learning_rate": 2.0600960135216463e-06,
+      "loss": 0.4041,
+      "step": 3050
+    },
+    {
+      "epoch": 12.24,
+      "grad_norm": 1.746474027633667,
+      "learning_rate": 2.003852089915548e-06,
+      "loss": 0.5115,
+      "step": 3060
+    },
+    {
+      "epoch": 12.28,
+      "grad_norm": 2.731269121170044,
+      "learning_rate": 1.9483010013976766e-06,
+      "loss": 0.4459,
+      "step": 3070
+    },
+    {
+      "epoch": 12.32,
+      "grad_norm": 1.9771169424057007,
+      "learning_rate": 1.8934475612536019e-06,
+      "loss": 0.3677,
+      "step": 3080
+    },
+    {
+      "epoch": 12.36,
+      "grad_norm": 3.043891191482544,
+      "learning_rate": 1.8392965223203707e-06,
+      "loss": 0.4353,
+      "step": 3090
+    },
+    {
+      "epoch": 12.4,
+      "grad_norm": 3.361074447631836,
+      "learning_rate": 1.7858525765747047e-06,
+      "loss": 0.4578,
+      "step": 3100
+    },
+    {
+      "epoch": 12.44,
+      "grad_norm": 3.8681182861328125,
+      "learning_rate": 1.7331203547264452e-06,
+      "loss": 0.4057,
+      "step": 3110
+    },
+    {
+      "epoch": 12.48,
+      "grad_norm": 2.581637382507324,
+      "learning_rate": 1.6811044258173425e-06,
+      "loss": 0.4532,
+      "step": 3120
+    },
+    {
+      "epoch": 12.52,
+      "grad_norm": 2.8616292476654053,
+      "learning_rate": 1.629809296825139e-06,
+      "loss": 0.4551,
+      "step": 3130
+    },
+    {
+      "epoch": 12.56,
+      "grad_norm": 2.4617111682891846,
+      "learning_rate": 1.579239412273078e-06,
+      "loss": 0.4388,
+      "step": 3140
+    },
+    {
+      "epoch": 12.6,
+      "grad_norm": 14.222563743591309,
+      "learning_rate": 1.5293991538447882e-06,
+      "loss": 0.412,
+      "step": 3150
+    },
+    {
+      "epoch": 12.64,
+      "grad_norm": 3.5921924114227295,
+      "learning_rate": 1.4802928400046457e-06,
+      "loss": 0.4517,
+      "step": 3160
+    },
+    {
+      "epoch": 12.68,
+      "grad_norm": 2.4046990871429443,
+      "learning_rate": 1.4319247256235713e-06,
+      "loss": 0.4893,
+      "step": 3170
+    },
+    {
+      "epoch": 12.72,
+      "grad_norm": 2.5496039390563965,
+      "learning_rate": 1.3842990016103886e-06,
+      "loss": 0.4305,
+      "step": 3180
+    },
+    {
+      "epoch": 12.76,
+      "grad_norm": 2.3980159759521484,
+      "learning_rate": 1.3374197945486833e-06,
+      "loss": 0.3833,
+      "step": 3190
+    },
+    {
+      "epoch": 12.8,
+      "grad_norm": 1.515519142150879,
+      "learning_rate": 1.2912911663392468e-06,
+      "loss": 0.4513,
+      "step": 3200
+    },
+    {
+      "epoch": 12.84,
+      "grad_norm": 2.939988136291504,
+      "learning_rate": 1.245917113848144e-06,
+      "loss": 0.4712,
+      "step": 3210
+    },
+    {
+      "epoch": 12.88,
+      "grad_norm": 1.846997857093811,
+      "learning_rate": 1.2013015685603813e-06,
+      "loss": 0.4789,
+      "step": 3220
+    },
+    {
+      "epoch": 12.92,
+      "grad_norm": 2.960897445678711,
+      "learning_rate": 1.1574483962392768e-06,
+      "loss": 0.4128,
+      "step": 3230
+    },
+    {
+      "epoch": 12.96,
+      "grad_norm": 1.8638381958007812,
+      "learning_rate": 1.114361396591498e-06,
+      "loss": 0.4949,
+      "step": 3240
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 2.134097099304199,
+      "learning_rate": 1.0720443029378303e-06,
+      "loss": 0.4167,
+      "step": 3250
+    },
+    {
+      "epoch": 13.04,
+      "grad_norm": 2.2456820011138916,
+      "learning_rate": 1.0305007818897006e-06,
+      "loss": 0.4483,
+      "step": 3260
+    },
+    {
+      "epoch": 13.08,
+      "grad_norm": 9.045520782470703,
+      "learning_rate": 9.897344330314862e-07,
+      "loss": 0.454,
+      "step": 3270
+    },
+    {
+      "epoch": 13.12,
+      "grad_norm": 2.646930694580078,
+      "learning_rate": 9.497487886086132e-07,
+      "loss": 0.4438,
+      "step": 3280
+    },
+    {
+      "epoch": 13.16,
+      "grad_norm": 4.203260898590088,
+      "learning_rate": 9.105473132215126e-07,
+      "loss": 0.3904,
+      "step": 3290
+    },
+    {
+      "epoch": 13.2,
+      "grad_norm": 3.177109479904175,
+      "learning_rate": 8.721334035254203e-07,
+      "loss": 0.4128,
+      "step": 3300
+    },
+    {
+      "epoch": 13.24,
+      "grad_norm": 1.825671911239624,
+      "learning_rate": 8.345103879360695e-07,
+      "loss": 0.4479,
+      "step": 3310
+    },
+    {
+      "epoch": 13.28,
+      "grad_norm": 3.2267651557922363,
+      "learning_rate": 7.976815263412963e-07,
+      "loss": 0.3928,
+      "step": 3320
+    },
+    {
+      "epoch": 13.32,
+      "grad_norm": 3.6811180114746094,
+      "learning_rate": 7.616500098185908e-07,
+      "loss": 0.4163,
+      "step": 3330
+    },
+    {
+      "epoch": 13.36,
+      "grad_norm": 3.258467435836792,
+      "learning_rate": 7.264189603585892e-07,
+      "loss": 0.4186,
+      "step": 3340
+    },
+    {
+      "epoch": 13.4,
+      "grad_norm": 1.861440658569336,
+      "learning_rate": 6.919914305945774e-07,
+      "loss": 0.4416,
+      "step": 3350
+    },
+    {
+      "epoch": 13.44,
+      "grad_norm": 3.303765296936035,
+      "learning_rate": 6.58370403537989e-07,
+      "loss": 0.3958,
+      "step": 3360
+    },
+    {
+      "epoch": 13.48,
+      "grad_norm": 1.9865639209747314,
+      "learning_rate": 6.255587923199313e-07,
+      "loss": 0.4424,
+      "step": 3370
+    },
+    {
+      "epoch": 13.52,
+      "grad_norm": 2.3877573013305664,
+      "learning_rate": 5.935594399387856e-07,
+      "loss": 0.4778,
+      "step": 3380
+    },
+    {
+      "epoch": 13.56,
+      "grad_norm": 2.6566929817199707,
+      "learning_rate": 5.623751190138682e-07,
+      "loss": 0.4045,
+      "step": 3390
+    },
+    {
+      "epoch": 13.6,
+      "grad_norm": 4.0699052810668945,
+      "learning_rate": 5.320085315451862e-07,
+      "loss": 0.4275,
+      "step": 3400
+    },
+    {
+      "epoch": 13.64,
+      "grad_norm": 4.235281944274902,
+      "learning_rate": 5.024623086793323e-07,
+      "loss": 0.4346,
+      "step": 3410
+    },
+    {
+      "epoch": 13.68,
+      "grad_norm": 1.7936820983886719,
+      "learning_rate": 4.737390104814954e-07,
+      "loss": 0.4343,
+      "step": 3420
+    },
+    {
+      "epoch": 13.72,
+      "grad_norm": 4.244424343109131,
+      "learning_rate": 4.458411257136486e-07,
+      "loss": 0.4355,
+      "step": 3430
+    },
+    {
+      "epoch": 13.76,
+      "grad_norm": 3.0799527168273926,
+      "learning_rate": 4.1877107161890416e-07,
+      "loss": 0.4407,
+      "step": 3440
+    },
+    {
+      "epoch": 13.8,
+      "grad_norm": 3.4574971199035645,
+      "learning_rate": 3.9253119371206684e-07,
+      "loss": 0.4369,
+      "step": 3450
+    },
+    {
+      "epoch": 13.84,
+      "grad_norm": 1.3924965858459473,
+      "learning_rate": 3.671237655764104e-07,
+      "loss": 0.4823,
+      "step": 3460
+    },
+    {
+      "epoch": 13.88,
+      "grad_norm": 1.9487115144729614,
+      "learning_rate": 3.4255098866667114e-07,
+      "loss": 0.4566,
+      "step": 3470
+    },
+    {
+      "epoch": 13.92,
+      "grad_norm": 2.6502346992492676,
+      "learning_rate": 3.188149921183115e-07,
+      "loss": 0.4824,
+      "step": 3480
+    },
+    {
+      "epoch": 13.96,
+      "grad_norm": 3.1728081703186035,
+      "learning_rate": 2.959178325630296e-07,
+      "loss": 0.3983,
+      "step": 3490
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 2.273251533508301,
+      "learning_rate": 2.7386149395056463e-07,
+      "loss": 0.4541,
+      "step": 3500
+    },
+    {
+      "epoch": 14.04,
+      "grad_norm": 2.2681076526641846,
+      "learning_rate": 2.526478873767946e-07,
+      "loss": 0.4667,
+      "step": 3510
+    },
+    {
+      "epoch": 14.08,
+      "grad_norm": 2.3255577087402344,
+      "learning_rate": 2.322788509181484e-07,
+      "loss": 0.441,
+      "step": 3520
+    },
+    {
+      "epoch": 14.12,
+      "grad_norm": 1.8558521270751953,
+      "learning_rate": 2.1275614947233624e-07,
+      "loss": 0.4294,
+      "step": 3530
+    },
+    {
+      "epoch": 14.16,
+      "grad_norm": 3.000098943710327,
+      "learning_rate": 1.9408147460544203e-07,
+      "loss": 0.4245,
+      "step": 3540
+    },
+    {
+      "epoch": 14.2,
+      "grad_norm": 3.5599377155303955,
+      "learning_rate": 1.7625644440534384e-07,
+      "loss": 0.3884,
+      "step": 3550
+    },
+    {
+      "epoch": 14.24,
+      "grad_norm": 3.208889961242676,
+      "learning_rate": 1.5928260334151847e-07,
+      "loss": 0.4434,
+      "step": 3560
+    },
+    {
+      "epoch": 14.28,
+      "grad_norm": 3.4625139236450195,
+      "learning_rate": 1.4316142213121386e-07,
+      "loss": 0.4474,
+      "step": 3570
+    },
+    {
+      "epoch": 14.32,
+      "grad_norm": 4.883245944976807,
+      "learning_rate": 1.2789429761202565e-07,
+      "loss": 0.3927,
+      "step": 3580
+    },
+    {
+      "epoch": 14.36,
+      "grad_norm": 4.052615165710449,
+      "learning_rate": 1.134825526208605e-07,
+      "loss": 0.3967,
+      "step": 3590
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 1.4959968328475952,
+      "learning_rate": 9.992743587931674e-08,
+      "loss": 0.4807,
+      "step": 3600
+    },
+    {
+      "epoch": 14.44,
+      "grad_norm": 2.757499933242798,
+      "learning_rate": 8.723012188549318e-08,
+      "loss": 0.4506,
+      "step": 3610
+    },
+    {
+      "epoch": 14.48,
+      "grad_norm": 2.0996549129486084,
+      "learning_rate": 7.539171081221597e-08,
+      "loss": 0.3934,
+      "step": 3620
+    },
+    {
+      "epoch": 14.52,
+      "grad_norm": 2.4962105751037598,
+      "learning_rate": 6.44132284117216e-08,
+      "loss": 0.4549,
+      "step": 3630
+    },
+    {
+      "epoch": 14.56,
+      "grad_norm": 1.9287108182907104,
+      "learning_rate": 5.429562592677018e-08,
+      "loss": 0.4208,
+      "step": 3640
+    },
+    {
+      "epoch": 14.6,
+      "grad_norm": 4.647493362426758,
+      "learning_rate": 4.503978000823028e-08,
+      "loss": 0.4376,
+      "step": 3650
+    },
+    {
+      "epoch": 14.64,
+      "grad_norm": 1.24240243434906,
+      "learning_rate": 3.6646492639118567e-08,
+      "loss": 0.4562,
+      "step": 3660
+    },
+    {
+      "epoch": 14.68,
+      "grad_norm": 2.359651565551758,
+      "learning_rate": 2.911649106511316e-08,
+      "loss": 0.42,
+      "step": 3670
+    },
+    {
+      "epoch": 14.72,
+      "grad_norm": 2.7108898162841797,
+      "learning_rate": 2.2450427731534052e-08,
+      "loss": 0.4489,
+      "step": 3680
+    },
+    {
+      "epoch": 14.76,
+      "grad_norm": 1.4741981029510498,
+      "learning_rate": 1.664888022682165e-08,
+      "loss": 0.4532,
+      "step": 3690
+    },
+    {
+      "epoch": 14.8,
+      "grad_norm": 1.676648497581482,
+      "learning_rate": 1.1712351232480157e-08,
+      "loss": 0.4375,
+      "step": 3700
+    },
+    {
+      "epoch": 14.84,
+      "grad_norm": 1.2909196615219116,
+      "learning_rate": 7.641268479531283e-09,
+      "loss": 0.4574,
+      "step": 3710
+    },
+    {
+      "epoch": 14.88,
+      "grad_norm": 3.2380595207214355,
+      "learning_rate": 4.435984711446128e-09,
+      "loss": 0.3538,
+      "step": 3720
+    },
+    {
+      "epoch": 14.92,
+      "grad_norm": 1.9485658407211304,
+      "learning_rate": 2.0967776535851802e-09,
+      "loss": 0.3884,
+      "step": 3730
+    },
+    {
+      "epoch": 14.96,
+      "grad_norm": 2.0014994144439697,
+      "learning_rate": 6.238499891353389e-10,
+      "loss": 0.4266,
+      "step": 3740
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 2.8281970024108887,
+      "learning_rate": 1.7329341542859922e-11,
+      "loss": 0.4743,
+      "step": 3750
+    },
+    {
+      "epoch": 15.0,
+      "step": 3750,
+      "total_flos": 8.815859931388314e+16,
+      "train_loss": 0.5605581335703532,
+      "train_runtime": 6084.0454,
+      "train_samples_per_second": 1.233,
+      "train_steps_per_second": 0.616
     }
   ],
   "logging_steps": 10,
-  "max_steps": 100,
+  "max_steps": 3750,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 15,
   "save_steps": 10000,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -105,7 +2660,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2315448779538432.0,
+  "total_flos": 8.815859931388314e+16,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:92129491714b846e96a93f7fce47a6f687ad2d877c101d404ffc46284b06dfda
+oid sha256:a7d13224a20770bb9bc99479f427296c451401a0aff0cc40e82e1b00254a2839
 size 5688