Instructions to use roonbug/mw4gx9uu with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use roonbug/mw4gx9uu with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="roonbug/mw4gx9uu")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("roonbug/mw4gx9uu")
model = AutoModelForImageTextToText.from_pretrained("roonbug/mw4gx9uu")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use roonbug/mw4gx9uu with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "roonbug/mw4gx9uu"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "roonbug/mw4gx9uu",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/roonbug/mw4gx9uu

SGLang

How to use roonbug/mw4gx9uu with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "roonbug/mw4gx9uu" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "roonbug/mw4gx9uu",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "roonbug/mw4gx9uu" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "roonbug/mw4gx9uu",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use roonbug/mw4gx9uu with Docker Model Runner:
```
docker model run hf.co/roonbug/mw4gx9uu
```

roonbug commited on May 3

Commit

0cfa078

verified ·

1 Parent(s): 0ce69f4

Upload folder using huggingface_hub

Browse files

Files changed (20) hide show

.gitattributes +1 -0
checkpoint-2500/added_tokens.json +3 -0
checkpoint-2500/chat_template.jinja +47 -0
checkpoint-2500/config.json +110 -0
checkpoint-2500/generation_config.json +13 -0
checkpoint-2500/model-00001-of-00005.safetensors +3 -0
checkpoint-2500/model-00002-of-00005.safetensors +3 -0
checkpoint-2500/model-00003-of-00005.safetensors +3 -0
checkpoint-2500/model-00004-of-00005.safetensors +3 -0
checkpoint-2500/model-00005-of-00005.safetensors +3 -0
checkpoint-2500/model.safetensors.index.json +0 -0
checkpoint-2500/optimizer.pt +3 -0
checkpoint-2500/rng_state.pth +3 -0
checkpoint-2500/scheduler.pt +3 -0
checkpoint-2500/special_tokens_map.json +33 -0
checkpoint-2500/tokenizer.json +3 -0
checkpoint-2500/tokenizer.model +3 -0
checkpoint-2500/tokenizer_config.json +0 -0
checkpoint-2500/trainer_state.json +3084 -0
checkpoint-2500/training_args.bin +3 -0

.gitattributes CHANGED Viewed

@@ -37,3 +37,4 @@ checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoint-1500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoint-2000/tokenizer.json filter=lfs diff=lfs merge=lfs -text

 checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoint-1500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoint-2000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-2500/tokenizer.json filter=lfs diff=lfs merge=lfs -text

checkpoint-2500/added_tokens.json ADDED Viewed

	@@ -0,0 +1,3 @@

+{
+  "<image_soft_token>": 262144
+}

checkpoint-2500/chat_template.jinja ADDED Viewed

	@@ -0,0 +1,47 @@

+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{'<start_of_turn>model
+'}}
+{%- endif -%}

checkpoint-2500/config.json ADDED Viewed

	@@ -0,0 +1,110 @@

+{
+  "architectures": [
+    "Gemma3ForConditionalGeneration"
+  ],
+  "boi_token_index": 255999,
+  "bos_token_id": 2,
+  "dtype": "bfloat16",
+  "eoi_token_index": 256000,
+  "eos_token_id": 1,
+  "image_token_index": 262144,
+  "initializer_range": 0.02,
+  "mm_tokens_per_image": 256,
+  "model_type": "gemma3",
+  "pad_token_id": 0,
+  "text_config": {
+    "_sliding_window_pattern": 6,
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attn_logit_softcapping": null,
+    "dtype": "bfloat16",
+    "final_logit_softcapping": null,
+    "head_dim": 256,
+    "hidden_activation": "gelu_pytorch_tanh",
+    "hidden_size": 3840,
+    "initializer_range": 0.02,
+    "intermediate_size": 15360,
+    "layer_types": [
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "sliding_attention",
+      "full_attention"
+    ],
+    "max_position_embeddings": 131072,
+    "model_type": "gemma3_text",
+    "num_attention_heads": 16,
+    "num_hidden_layers": 48,
+    "num_key_value_heads": 8,
+    "query_pre_attn_scalar": 256,
+    "rms_norm_eps": 1e-06,
+    "rope_local_base_freq": 10000.0,
+    "rope_scaling": {
+      "factor": 8.0,
+      "rope_type": "linear"
+    },
+    "rope_theta": 1000000.0,
+    "sliding_window": 1024,
+    "use_cache": true,
+    "vocab_size": 262208
+  },
+  "transformers_version": "4.56.1",
+  "vision_config": {
+    "attention_dropout": 0.0,
+    "dtype": "bfloat16",
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 1152,
+    "image_size": 896,
+    "intermediate_size": 4304,
+    "layer_norm_eps": 1e-06,
+    "model_type": "siglip_vision_model",
+    "num_attention_heads": 16,
+    "num_channels": 3,
+    "num_hidden_layers": 27,
+    "patch_size": 14,
+    "vision_use_head": false
+  }
+}

checkpoint-2500/generation_config.json ADDED Viewed

	@@ -0,0 +1,13 @@

+{
+  "bos_token_id": 2,
+  "do_sample": true,
+  "eos_token_id": [
+    1,
+    1,
+    106
+  ],
+  "pad_token_id": 0,
+  "top_k": 64,
+  "top_p": 0.95,
+  "transformers_version": "4.56.1"
+}

checkpoint-2500/model-00001-of-00005.safetensors ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:4847447e92599833e8dbaa3067cd201c3bb5c052efa91f11ba891e43234f7832
+size 4979902192

checkpoint-2500/model-00002-of-00005.safetensors ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:891bd54eed03cba9ee1e705533a02a8217fcc29f356e4a1f53e5fd0d178883ad
+size 4931296592

checkpoint-2500/model-00003-of-00005.safetensors ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:7cee411d9d57324e50ce064a192cc5a858276d508611b12fc599e0c9767112e0
+size 4931296656

checkpoint-2500/model-00004-of-00005.safetensors ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:2543d5aa67e0b4d65eaada3953806edb311e05c4fd404b6298b5002eb5ba691f
+size 4931296656

checkpoint-2500/model-00005-of-00005.safetensors ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:4f784176bb7b972e169e6b31120c3c2d99229c9d66ab8c0f941d4044f63da3b6
+size 4601000928

checkpoint-2500/model.safetensors.index.json ADDED Viewed

The diff for this file is too large to render. See raw diff

checkpoint-2500/optimizer.pt ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:5b1be9a8bb198629badaa7f435751cec8ba5d5b454c33238cf3bbbddb1df6fab
+size 7286558019

checkpoint-2500/rng_state.pth ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:8bbd07b04584d51483a537ebe86fdb5e065d175fef715f69db1afc368ee4945d
+size 14645

checkpoint-2500/scheduler.pt ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:401374f6391dea991900232c652375909f9d1e28f273a2842bccb16d538e61ce
+size 1465

checkpoint-2500/special_tokens_map.json ADDED Viewed

	@@ -0,0 +1,33 @@

+{
+  "boi_token": "<start_of_image>",
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eoi_token": "<end_of_image>",
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "image_token": "<image_soft_token>",
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-2500/tokenizer.json ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:4667f2089529e8e7657cfb6d1c19910ae71ff5f28aa7ab2ff2763330affad795
+size 33384568

checkpoint-2500/tokenizer.model ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:1299c11d7cf632ef3b4e11937501358ada021bbdf7c47638d13c0ee982f2e79c
+size 4689074

checkpoint-2500/tokenizer_config.json ADDED Viewed

The diff for this file is too large to render. See raw diff

checkpoint-2500/trainer_state.json ADDED Viewed

	@@ -0,0 +1,3084 @@

+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 100,
+  "global_step": 2500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 1.0092366490513087,
+      "epoch": 0.016,
+      "grad_norm": 74.5,
+      "learning_rate": 1.8e-07,
+      "loss": 18.8937,
+      "mean_token_accuracy": 0.7410404551774263,
+      "num_tokens": 338352.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.0230814347043633,
+      "epoch": 0.032,
+      "grad_norm": 76.0,
+      "learning_rate": 3.8e-07,
+      "loss": 18.9456,
+      "mean_token_accuracy": 0.7410476118326187,
+      "num_tokens": 671193.0,
+      "step": 20
+    },
+    {
+      "entropy": 1.012267287634313,
+      "epoch": 0.048,
+      "grad_norm": 47.0,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 18.671,
+      "mean_token_accuracy": 0.7421887803822755,
+      "num_tokens": 1001008.0,
+      "step": 30
+    },
+    {
+      "entropy": 1.0519745318219065,
+      "epoch": 0.064,
+      "grad_norm": 50.25,
+      "learning_rate": 7.8e-07,
+      "loss": 19.1892,
+      "mean_token_accuracy": 0.73549969419837,
+      "num_tokens": 1331208.0,
+      "step": 40
+    },
+    {
+      "entropy": 1.0362586732953787,
+      "epoch": 0.08,
+      "grad_norm": 60.5,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 18.5399,
+      "mean_token_accuracy": 0.7425773791968823,
+      "num_tokens": 1667532.0,
+      "step": 50
+    },
+    {
+      "entropy": 1.0199208352714777,
+      "epoch": 0.096,
+      "grad_norm": 47.0,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 18.0825,
+      "mean_token_accuracy": 0.7469310026615859,
+      "num_tokens": 2007176.0,
+      "step": 60
+    },
+    {
+      "entropy": 1.0730392780154943,
+      "epoch": 0.112,
+      "grad_norm": 40.25,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 18.5203,
+      "mean_token_accuracy": 0.7405710384249687,
+      "num_tokens": 2342259.0,
+      "step": 70
+    },
+    {
+      "entropy": 1.158516400679946,
+      "epoch": 0.128,
+      "grad_norm": 48.5,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 19.4751,
+      "mean_token_accuracy": 0.7302160415798425,
+      "num_tokens": 2663987.0,
+      "step": 80
+    },
+    {
+      "entropy": 1.103706269338727,
+      "epoch": 0.144,
+      "grad_norm": 34.0,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 18.236,
+      "mean_token_accuracy": 0.7424531534314156,
+      "num_tokens": 2997824.0,
+      "step": 90
+    },
+    {
+      "entropy": 1.1111123770475388,
+      "epoch": 0.16,
+      "grad_norm": 41.5,
+      "learning_rate": 1.98e-06,
+      "loss": 18.1903,
+      "mean_token_accuracy": 0.7447310462594032,
+      "num_tokens": 3330597.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.6746724898815155,
+      "eval_math_loss": 0.6892878413200378,
+      "eval_math_mean_token_accuracy": 0.8155246863365173,
+      "eval_math_num_tokens": 3330597.0,
+      "eval_math_runtime": 37.2985,
+      "eval_math_samples_per_second": 13.405,
+      "eval_math_steps_per_second": 3.351,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 1.1165525755882264,
+      "eval_physics_loss": 1.1379250288009644,
+      "eval_physics_mean_token_accuracy": 0.7427083716392517,
+      "eval_physics_num_tokens": 3330597.0,
+      "eval_physics_runtime": 42.3918,
+      "eval_physics_samples_per_second": 11.795,
+      "eval_physics_steps_per_second": 2.949,
+      "step": 100
+    },
+    {
+      "entropy": 1.1296000629663467,
+      "epoch": 0.176,
+      "grad_norm": 43.0,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 18.5006,
+      "mean_token_accuracy": 0.7395337466150522,
+      "num_tokens": 3658264.0,
+      "step": 110
+    },
+    {
+      "entropy": 1.1084317233413459,
+      "epoch": 0.192,
+      "grad_norm": 28.125,
+      "learning_rate": 2.38e-06,
+      "loss": 18.0132,
+      "mean_token_accuracy": 0.7432477563619614,
+      "num_tokens": 3995568.0,
+      "step": 120
+    },
+    {
+      "entropy": 1.0731435023248195,
+      "epoch": 0.208,
+      "grad_norm": 34.25,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 17.4019,
+      "mean_token_accuracy": 0.7507147442549467,
+      "num_tokens": 4321436.0,
+      "step": 130
+    },
+    {
+      "entropy": 1.084234836883843,
+      "epoch": 0.224,
+      "grad_norm": 28.0,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 17.5116,
+      "mean_token_accuracy": 0.747317761927843,
+      "num_tokens": 4648491.0,
+      "step": 140
+    },
+    {
+      "entropy": 1.0654050692915917,
+      "epoch": 0.24,
+      "grad_norm": 28.75,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 17.2926,
+      "mean_token_accuracy": 0.750689634680748,
+      "num_tokens": 4986175.0,
+      "step": 150
+    },
+    {
+      "entropy": 1.0740458451211452,
+      "epoch": 0.256,
+      "grad_norm": 30.75,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 17.2845,
+      "mean_token_accuracy": 0.7503778677433729,
+      "num_tokens": 5329320.0,
+      "step": 160
+    },
+    {
+      "entropy": 1.0785667825490237,
+      "epoch": 0.272,
+      "grad_norm": 31.5,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 17.3538,
+      "mean_token_accuracy": 0.7468075532466173,
+      "num_tokens": 5658796.0,
+      "step": 170
+    },
+    {
+      "entropy": 1.0575236398726702,
+      "epoch": 0.288,
+      "grad_norm": 27.0,
+      "learning_rate": 3.58e-06,
+      "loss": 17.2248,
+      "mean_token_accuracy": 0.7524373367428779,
+      "num_tokens": 5980201.0,
+      "step": 180
+    },
+    {
+      "entropy": 1.0599254176020623,
+      "epoch": 0.304,
+      "grad_norm": 33.75,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 17.0558,
+      "mean_token_accuracy": 0.7516172304749489,
+      "num_tokens": 6298858.0,
+      "step": 190
+    },
+    {
+      "entropy": 1.06340769007802,
+      "epoch": 0.32,
+      "grad_norm": 25.875,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 17.0226,
+      "mean_token_accuracy": 0.7516651555895806,
+      "num_tokens": 6622798.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.6770454683303833,
+      "eval_math_loss": 0.6920014023780823,
+      "eval_math_mean_token_accuracy": 0.8149451169967651,
+      "eval_math_num_tokens": 6622798.0,
+      "eval_math_runtime": 37.3199,
+      "eval_math_samples_per_second": 13.398,
+      "eval_math_steps_per_second": 3.349,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 1.0526363019943237,
+      "eval_physics_loss": 1.0609160661697388,
+      "eval_physics_mean_token_accuracy": 0.7532025098800659,
+      "eval_physics_num_tokens": 6622798.0,
+      "eval_physics_runtime": 42.3894,
+      "eval_physics_samples_per_second": 11.795,
+      "eval_physics_steps_per_second": 2.949,
+      "step": 200
+    },
+    {
+      "entropy": 1.0413700450211762,
+      "epoch": 0.336,
+      "grad_norm": 22.75,
+      "learning_rate": 4.18e-06,
+      "loss": 16.7861,
+      "mean_token_accuracy": 0.7550040531903506,
+      "num_tokens": 6953611.0,
+      "step": 210
+    },
+    {
+      "entropy": 1.0119973294436932,
+      "epoch": 0.352,
+      "grad_norm": 26.875,
+      "learning_rate": 4.38e-06,
+      "loss": 16.2682,
+      "mean_token_accuracy": 0.7610075753182173,
+      "num_tokens": 7290378.0,
+      "step": 220
+    },
+    {
+      "entropy": 1.01713064648211,
+      "epoch": 0.368,
+      "grad_norm": 28.75,
+      "learning_rate": 4.58e-06,
+      "loss": 16.3086,
+      "mean_token_accuracy": 0.7611279483884573,
+      "num_tokens": 7621082.0,
+      "step": 230
+    },
+    {
+      "entropy": 1.0563053917139769,
+      "epoch": 0.384,
+      "grad_norm": 27.125,
+      "learning_rate": 4.78e-06,
+      "loss": 17.0452,
+      "mean_token_accuracy": 0.7495943393558264,
+      "num_tokens": 7955570.0,
+      "step": 240
+    },
+    {
+      "entropy": 1.0173547353595496,
+      "epoch": 0.4,
+      "grad_norm": 29.5,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 16.2976,
+      "mean_token_accuracy": 0.7596513438969851,
+      "num_tokens": 8291049.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.984210986085236,
+      "epoch": 0.416,
+      "grad_norm": 27.0,
+      "learning_rate": 5.18e-06,
+      "loss": 15.9194,
+      "mean_token_accuracy": 0.7646323684602976,
+      "num_tokens": 8627310.0,
+      "step": 260
+    },
+    {
+      "entropy": 1.0042537026107312,
+      "epoch": 0.432,
+      "grad_norm": 29.625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 16.1695,
+      "mean_token_accuracy": 0.7608209282159806,
+      "num_tokens": 8958371.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.9618185676634312,
+      "epoch": 0.448,
+      "grad_norm": 27.875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 15.4588,
+      "mean_token_accuracy": 0.7696380686014891,
+      "num_tokens": 9294388.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.9816760478541255,
+      "epoch": 0.464,
+      "grad_norm": 34.0,
+      "learning_rate": 5.78e-06,
+      "loss": 15.6523,
+      "mean_token_accuracy": 0.7662083100527525,
+      "num_tokens": 9620366.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.9721551053225994,
+      "epoch": 0.48,
+      "grad_norm": 30.625,
+      "learning_rate": 5.98e-06,
+      "loss": 15.7877,
+      "mean_token_accuracy": 0.7654597967863083,
+      "num_tokens": 9955431.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.6786130204200744,
+      "eval_math_loss": 0.6952713131904602,
+      "eval_math_mean_token_accuracy": 0.8145794682502746,
+      "eval_math_num_tokens": 9955431.0,
+      "eval_math_runtime": 37.3306,
+      "eval_math_samples_per_second": 13.394,
+      "eval_math_steps_per_second": 3.348,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 1.001317280292511,
+      "eval_physics_loss": 1.0042355060577393,
+      "eval_physics_mean_token_accuracy": 0.7617036099433899,
+      "eval_physics_num_tokens": 9955431.0,
+      "eval_physics_runtime": 42.4154,
+      "eval_physics_samples_per_second": 11.788,
+      "eval_physics_steps_per_second": 2.947,
+      "step": 300
+    },
+    {
+      "entropy": 0.998606894724071,
+      "epoch": 0.496,
+      "grad_norm": 27.0,
+      "learning_rate": 6.18e-06,
+      "loss": 15.9637,
+      "mean_token_accuracy": 0.7627440579235554,
+      "num_tokens": 10289883.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.9797808565199375,
+      "epoch": 0.512,
+      "grad_norm": 31.5,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 15.7986,
+      "mean_token_accuracy": 0.7650973781943321,
+      "num_tokens": 10619468.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.9935696642845869,
+      "epoch": 0.528,
+      "grad_norm": 26.625,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 16.0109,
+      "mean_token_accuracy": 0.7615033354610204,
+      "num_tokens": 10944904.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.9842395279556513,
+      "epoch": 0.544,
+      "grad_norm": 30.375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 15.7481,
+      "mean_token_accuracy": 0.7652762945741415,
+      "num_tokens": 11266409.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.9608212519437075,
+      "epoch": 0.56,
+      "grad_norm": 24.5,
+      "learning_rate": 6.98e-06,
+      "loss": 15.4249,
+      "mean_token_accuracy": 0.769417317211628,
+      "num_tokens": 11605544.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.935908448882401,
+      "epoch": 0.576,
+      "grad_norm": 27.375,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 15.0616,
+      "mean_token_accuracy": 0.773730392754078,
+      "num_tokens": 11931715.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.9477673010900617,
+      "epoch": 0.592,
+      "grad_norm": 27.75,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 15.2839,
+      "mean_token_accuracy": 0.7681151241064071,
+      "num_tokens": 12257243.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.9456456508487463,
+      "epoch": 0.608,
+      "grad_norm": 26.875,
+      "learning_rate": 7.58e-06,
+      "loss": 15.0726,
+      "mean_token_accuracy": 0.7709438435733318,
+      "num_tokens": 12583154.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.9472203815355897,
+      "epoch": 0.624,
+      "grad_norm": 26.0,
+      "learning_rate": 7.78e-06,
+      "loss": 15.2406,
+      "mean_token_accuracy": 0.7696458593010902,
+      "num_tokens": 12905392.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.955901001021266,
+      "epoch": 0.64,
+      "grad_norm": 27.0,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 15.4306,
+      "mean_token_accuracy": 0.7679449453949928,
+      "num_tokens": 13232198.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.6780883803367614,
+      "eval_math_loss": 0.699813723564148,
+      "eval_math_mean_token_accuracy": 0.8138141121864318,
+      "eval_math_num_tokens": 13232198.0,
+      "eval_math_runtime": 37.2424,
+      "eval_math_samples_per_second": 13.426,
+      "eval_math_steps_per_second": 3.356,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.9454385538101197,
+      "eval_physics_loss": 0.954737663269043,
+      "eval_physics_mean_token_accuracy": 0.7692720980644226,
+      "eval_physics_num_tokens": 13232198.0,
+      "eval_physics_runtime": 42.3209,
+      "eval_physics_samples_per_second": 11.815,
+      "eval_physics_steps_per_second": 2.954,
+      "step": 400
+    },
+    {
+      "entropy": 0.9476992039009928,
+      "epoch": 0.656,
+      "grad_norm": 25.875,
+      "learning_rate": 8.18e-06,
+      "loss": 15.0903,
+      "mean_token_accuracy": 0.7713554214686156,
+      "num_tokens": 13575902.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.9169062672182917,
+      "epoch": 0.672,
+      "grad_norm": 31.0,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 14.8261,
+      "mean_token_accuracy": 0.7750746656209231,
+      "num_tokens": 13895997.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.9244599049910903,
+      "epoch": 0.688,
+      "grad_norm": 27.75,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 14.8502,
+      "mean_token_accuracy": 0.7752744313329458,
+      "num_tokens": 14234888.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.933361453562975,
+      "epoch": 0.704,
+      "grad_norm": 33.75,
+      "learning_rate": 8.78e-06,
+      "loss": 14.9746,
+      "mean_token_accuracy": 0.7716180372983217,
+      "num_tokens": 14567908.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.9235167052596808,
+      "epoch": 0.72,
+      "grad_norm": 32.75,
+      "learning_rate": 8.98e-06,
+      "loss": 14.8192,
+      "mean_token_accuracy": 0.7735234662890434,
+      "num_tokens": 14882927.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.9158632257953286,
+      "epoch": 0.736,
+      "grad_norm": 31.625,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 14.7253,
+      "mean_token_accuracy": 0.7745446749031544,
+      "num_tokens": 15217342.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.9053685218095779,
+      "epoch": 0.752,
+      "grad_norm": 27.25,
+      "learning_rate": 9.38e-06,
+      "loss": 14.4754,
+      "mean_token_accuracy": 0.7773572497069836,
+      "num_tokens": 15547710.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.939404590241611,
+      "epoch": 0.768,
+      "grad_norm": 30.625,
+      "learning_rate": 9.58e-06,
+      "loss": 15.169,
+      "mean_token_accuracy": 0.7681807715445756,
+      "num_tokens": 15877177.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.8850956622511148,
+      "epoch": 0.784,
+      "grad_norm": 24.5,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 14.1945,
+      "mean_token_accuracy": 0.7821484938263893,
+      "num_tokens": 16219640.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.899562518298626,
+      "epoch": 0.8,
+      "grad_norm": 31.0,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 14.4725,
+      "mean_token_accuracy": 0.7781291462481021,
+      "num_tokens": 16548261.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.6752088174819947,
+      "eval_math_loss": 0.7044452428817749,
+      "eval_math_mean_token_accuracy": 0.8128278303146362,
+      "eval_math_num_tokens": 16548261.0,
+      "eval_math_runtime": 37.3074,
+      "eval_math_samples_per_second": 13.402,
+      "eval_math_steps_per_second": 3.351,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.9063508882522583,
+      "eval_physics_loss": 0.913874089717865,
+      "eval_physics_mean_token_accuracy": 0.7756751170158386,
+      "eval_physics_num_tokens": 16548261.0,
+      "eval_physics_runtime": 42.2317,
+      "eval_physics_samples_per_second": 11.839,
+      "eval_physics_steps_per_second": 2.96,
+      "step": 500
+    },
+    {
+      "entropy": 0.8967189636081457,
+      "epoch": 0.816,
+      "grad_norm": 27.75,
+      "learning_rate": 1.018e-05,
+      "loss": 14.4259,
+      "mean_token_accuracy": 0.776664936169982,
+      "num_tokens": 16870533.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.8994548883289099,
+      "epoch": 0.832,
+      "grad_norm": 27.5,
+      "learning_rate": 1.038e-05,
+      "loss": 14.4201,
+      "mean_token_accuracy": 0.7781281162053346,
+      "num_tokens": 17196371.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.8786382066085935,
+      "epoch": 0.848,
+      "grad_norm": 23.875,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 14.1629,
+      "mean_token_accuracy": 0.7804178979247809,
+      "num_tokens": 17527066.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.9056067479774356,
+      "epoch": 0.864,
+      "grad_norm": 27.75,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 14.4586,
+      "mean_token_accuracy": 0.7776577699929476,
+      "num_tokens": 17859605.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.9013458237051963,
+      "epoch": 0.88,
+      "grad_norm": 30.75,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 14.4507,
+      "mean_token_accuracy": 0.7773088499903679,
+      "num_tokens": 18187315.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.8363397585228085,
+      "epoch": 0.896,
+      "grad_norm": 24.875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 13.4286,
+      "mean_token_accuracy": 0.7901293423026801,
+      "num_tokens": 18536790.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.8889346038922668,
+      "epoch": 0.912,
+      "grad_norm": 27.625,
+      "learning_rate": 1.138e-05,
+      "loss": 14.2651,
+      "mean_token_accuracy": 0.778273418545723,
+      "num_tokens": 18869344.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.8685549266636372,
+      "epoch": 0.928,
+      "grad_norm": 28.375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 14.0196,
+      "mean_token_accuracy": 0.7813166368752718,
+      "num_tokens": 19199834.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.8552317110821605,
+      "epoch": 0.944,
+      "grad_norm": 32.75,
+      "learning_rate": 1.178e-05,
+      "loss": 13.7437,
+      "mean_token_accuracy": 0.7851793665438891,
+      "num_tokens": 19525385.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.8605958197265864,
+      "epoch": 0.96,
+      "grad_norm": 35.75,
+      "learning_rate": 1.198e-05,
+      "loss": 13.8362,
+      "mean_token_accuracy": 0.7834600981324911,
+      "num_tokens": 19856973.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.6771966469287872,
+      "eval_math_loss": 0.7089411020278931,
+      "eval_math_mean_token_accuracy": 0.8117777819633484,
+      "eval_math_num_tokens": 19856973.0,
+      "eval_math_runtime": 37.3206,
+      "eval_math_samples_per_second": 13.397,
+      "eval_math_steps_per_second": 3.349,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.8764889559745789,
+      "eval_physics_loss": 0.8784005641937256,
+      "eval_physics_mean_token_accuracy": 0.7810807003974914,
+      "eval_physics_num_tokens": 19856973.0,
+      "eval_physics_runtime": 42.4096,
+      "eval_physics_samples_per_second": 11.79,
+      "eval_physics_steps_per_second": 2.947,
+      "step": 600
+    },
+    {
+      "entropy": 0.8510318692773581,
+      "epoch": 0.976,
+      "grad_norm": 28.875,
+      "learning_rate": 1.218e-05,
+      "loss": 13.7003,
+      "mean_token_accuracy": 0.7870899338275195,
+      "num_tokens": 20190600.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.8734505966305732,
+      "epoch": 0.992,
+      "grad_norm": 25.625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 13.9297,
+      "mean_token_accuracy": 0.7843606945127248,
+      "num_tokens": 20518014.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.8346759399399162,
+      "epoch": 1.008,
+      "grad_norm": 26.625,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 13.3082,
+      "mean_token_accuracy": 0.7894357409328222,
+      "num_tokens": 20859215.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.8322532791644335,
+      "epoch": 1.024,
+      "grad_norm": 26.125,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 13.4039,
+      "mean_token_accuracy": 0.7888920601457358,
+      "num_tokens": 21186658.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.8312261745333671,
+      "epoch": 1.04,
+      "grad_norm": 24.25,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 13.3443,
+      "mean_token_accuracy": 0.7897948380559683,
+      "num_tokens": 21517237.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.8096720686182379,
+      "epoch": 1.056,
+      "grad_norm": 25.625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 13.0267,
+      "mean_token_accuracy": 0.7933571342378855,
+      "num_tokens": 21852528.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.828827659972012,
+      "epoch": 1.072,
+      "grad_norm": 23.125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 13.2583,
+      "mean_token_accuracy": 0.7900710258632898,
+      "num_tokens": 22192380.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.8289778597652913,
+      "epoch": 1.088,
+      "grad_norm": 28.875,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 13.2961,
+      "mean_token_accuracy": 0.7894763175398112,
+      "num_tokens": 22517971.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.8111658392474055,
+      "epoch": 1.104,
+      "grad_norm": 24.625,
+      "learning_rate": 1.378e-05,
+      "loss": 13.068,
+      "mean_token_accuracy": 0.7919723801314831,
+      "num_tokens": 22848694.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.819352557323873,
+      "epoch": 1.12,
+      "grad_norm": 27.0,
+      "learning_rate": 1.398e-05,
+      "loss": 13.1764,
+      "mean_token_accuracy": 0.790574399009347,
+      "num_tokens": 23183889.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6765684578418731,
+      "eval_math_loss": 0.7148844003677368,
+      "eval_math_mean_token_accuracy": 0.8105084261894226,
+      "eval_math_num_tokens": 23183889.0,
+      "eval_math_runtime": 37.3056,
+      "eval_math_samples_per_second": 13.403,
+      "eval_math_steps_per_second": 3.351,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.8618767859935761,
+      "eval_physics_loss": 0.851262092590332,
+      "eval_physics_mean_token_accuracy": 0.7858785772323609,
+      "eval_physics_num_tokens": 23183889.0,
+      "eval_physics_runtime": 42.3964,
+      "eval_physics_samples_per_second": 11.793,
+      "eval_physics_steps_per_second": 2.948,
+      "step": 700
+    },
+    {
+      "entropy": 0.7947209561243653,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 24.0,
+      "learning_rate": 1.418e-05,
+      "loss": 12.7353,
+      "mean_token_accuracy": 0.7966287981718778,
+      "num_tokens": 23523903.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.8054657407104969,
+      "epoch": 1.152,
+      "grad_norm": 26.375,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 12.9867,
+      "mean_token_accuracy": 0.7943034041672945,
+      "num_tokens": 23848848.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.8241605624556542,
+      "epoch": 1.168,
+      "grad_norm": 30.625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 13.2719,
+      "mean_token_accuracy": 0.7889104891568423,
+      "num_tokens": 24174589.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.8090734414756298,
+      "epoch": 1.184,
+      "grad_norm": 27.375,
+      "learning_rate": 1.478e-05,
+      "loss": 12.9551,
+      "mean_token_accuracy": 0.7938775520771741,
+      "num_tokens": 24495981.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.7889803951606155,
+      "epoch": 1.2,
+      "grad_norm": 26.25,
+      "learning_rate": 1.498e-05,
+      "loss": 12.7263,
+      "mean_token_accuracy": 0.7964807048439979,
+      "num_tokens": 24826245.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.8167023938149214,
+      "epoch": 1.216,
+      "grad_norm": 31.625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 13.0451,
+      "mean_token_accuracy": 0.7903591353446245,
+      "num_tokens": 25147775.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.789948632195592,
+      "epoch": 1.232,
+      "grad_norm": 26.0,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 12.7637,
+      "mean_token_accuracy": 0.7966345235705375,
+      "num_tokens": 25480033.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.8032306229695678,
+      "epoch": 1.248,
+      "grad_norm": 25.375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 12.7939,
+      "mean_token_accuracy": 0.7953033585101366,
+      "num_tokens": 25811978.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.7868311481550336,
+      "epoch": 1.264,
+      "grad_norm": 30.5,
+      "learning_rate": 1.578e-05,
+      "loss": 12.6982,
+      "mean_token_accuracy": 0.7961065553128719,
+      "num_tokens": 26146299.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.8258139120414854,
+      "epoch": 1.28,
+      "grad_norm": 29.125,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 13.1783,
+      "mean_token_accuracy": 0.7892586536705494,
+      "num_tokens": 26478207.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6665113575458527,
+      "eval_math_loss": 0.7234205603599548,
+      "eval_math_mean_token_accuracy": 0.809327778339386,
+      "eval_math_num_tokens": 26478207.0,
+      "eval_math_runtime": 37.3419,
+      "eval_math_samples_per_second": 13.39,
+      "eval_math_steps_per_second": 3.347,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.8228767182826996,
+      "eval_physics_loss": 0.8289657235145569,
+      "eval_physics_mean_token_accuracy": 0.7892685952186584,
+      "eval_physics_num_tokens": 26478207.0,
+      "eval_physics_runtime": 42.4475,
+      "eval_physics_samples_per_second": 11.779,
+      "eval_physics_steps_per_second": 2.945,
+      "step": 800
+    },
+    {
+      "entropy": 0.7805587112903595,
+      "epoch": 1.296,
+      "grad_norm": 27.75,
+      "learning_rate": 1.618e-05,
+      "loss": 12.5534,
+      "mean_token_accuracy": 0.7988621093332767,
+      "num_tokens": 26806477.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.8149383123964071,
+      "epoch": 1.312,
+      "grad_norm": 27.125,
+      "learning_rate": 1.638e-05,
+      "loss": 13.0917,
+      "mean_token_accuracy": 0.7913250483572483,
+      "num_tokens": 27137528.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.7897355943918228,
+      "epoch": 1.328,
+      "grad_norm": 30.375,
+      "learning_rate": 1.658e-05,
+      "loss": 12.7726,
+      "mean_token_accuracy": 0.7949048325419426,
+      "num_tokens": 27465859.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.807949285581708,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 34.75,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 12.9002,
+      "mean_token_accuracy": 0.7933759417384862,
+      "num_tokens": 27795568.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.8059511505067348,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 21.875,
+      "learning_rate": 1.698e-05,
+      "loss": 12.87,
+      "mean_token_accuracy": 0.793031033501029,
+      "num_tokens": 28129988.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.7793916024267673,
+      "epoch": 1.376,
+      "grad_norm": 27.0,
+      "learning_rate": 1.718e-05,
+      "loss": 12.617,
+      "mean_token_accuracy": 0.7958804558962583,
+      "num_tokens": 28465903.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.7730198312550783,
+      "epoch": 1.392,
+      "grad_norm": 32.25,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 12.3427,
+      "mean_token_accuracy": 0.8008152190595865,
+      "num_tokens": 28801554.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.776485905982554,
+      "epoch": 1.408,
+      "grad_norm": 24.75,
+      "learning_rate": 1.758e-05,
+      "loss": 12.4852,
+      "mean_token_accuracy": 0.7977169454097748,
+      "num_tokens": 29125114.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.800671405531466,
+      "epoch": 1.424,
+      "grad_norm": 24.125,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 12.8697,
+      "mean_token_accuracy": 0.7924031157046556,
+      "num_tokens": 29454334.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.7569738568738102,
+      "epoch": 1.44,
+      "grad_norm": 23.75,
+      "learning_rate": 1.798e-05,
+      "loss": 12.1405,
+      "mean_token_accuracy": 0.803036654368043,
+      "num_tokens": 29789253.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6468490760326385,
+      "eval_math_loss": 0.7321750521659851,
+      "eval_math_mean_token_accuracy": 0.8078626461029053,
+      "eval_math_num_tokens": 29789253.0,
+      "eval_math_runtime": 37.3449,
+      "eval_math_samples_per_second": 13.389,
+      "eval_math_steps_per_second": 3.347,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.7729090967178345,
+      "eval_physics_loss": 0.8055168390274048,
+      "eval_physics_mean_token_accuracy": 0.7930156898498535,
+      "eval_physics_num_tokens": 29789253.0,
+      "eval_physics_runtime": 42.7168,
+      "eval_physics_samples_per_second": 11.705,
+      "eval_physics_steps_per_second": 2.926,
+      "step": 900
+    },
+    {
+      "entropy": 0.7531193299219012,
+      "epoch": 1.456,
+      "grad_norm": 21.625,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 12.1313,
+      "mean_token_accuracy": 0.8043223068118095,
+      "num_tokens": 30128576.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.8030919276177884,
+      "epoch": 1.472,
+      "grad_norm": 26.875,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 12.9003,
+      "mean_token_accuracy": 0.7934082742780447,
+      "num_tokens": 30452719.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.7907818842679262,
+      "epoch": 1.488,
+      "grad_norm": 22.875,
+      "learning_rate": 1.858e-05,
+      "loss": 12.6202,
+      "mean_token_accuracy": 0.7969732403755188,
+      "num_tokens": 30782343.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.763646744005382,
+      "epoch": 1.504,
+      "grad_norm": 21.375,
+      "learning_rate": 1.878e-05,
+      "loss": 12.2876,
+      "mean_token_accuracy": 0.8009975593537092,
+      "num_tokens": 31113238.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.7834568763151765,
+      "epoch": 1.52,
+      "grad_norm": 26.875,
+      "learning_rate": 1.898e-05,
+      "loss": 12.5808,
+      "mean_token_accuracy": 0.7964470848441124,
+      "num_tokens": 31444589.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.7918407360091806,
+      "epoch": 1.536,
+      "grad_norm": 28.75,
+      "learning_rate": 1.918e-05,
+      "loss": 12.7438,
+      "mean_token_accuracy": 0.7937989573925733,
+      "num_tokens": 31774137.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.7553022712469101,
+      "epoch": 1.552,
+      "grad_norm": 24.875,
+      "learning_rate": 1.938e-05,
+      "loss": 12.1698,
+      "mean_token_accuracy": 0.8028277900069952,
+      "num_tokens": 32101776.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.7618934854865074,
+      "epoch": 1.568,
+      "grad_norm": 24.625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 12.194,
+      "mean_token_accuracy": 0.8020534928888082,
+      "num_tokens": 32428021.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.7852840103209019,
+      "epoch": 1.584,
+      "grad_norm": 23.875,
+      "learning_rate": 1.978e-05,
+      "loss": 12.6209,
+      "mean_token_accuracy": 0.793876089900732,
+      "num_tokens": 32752735.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.7698918517678977,
+      "epoch": 1.6,
+      "grad_norm": 23.125,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 12.4372,
+      "mean_token_accuracy": 0.7988904159516096,
+      "num_tokens": 33075822.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6642975335121155,
+      "eval_math_loss": 0.7357895970344543,
+      "eval_math_mean_token_accuracy": 0.8065607080459595,
+      "eval_math_num_tokens": 33075822.0,
+      "eval_math_runtime": 37.2859,
+      "eval_math_samples_per_second": 13.41,
+      "eval_math_steps_per_second": 3.352,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.7814281928539276,
+      "eval_physics_loss": 0.7896368503570557,
+      "eval_physics_mean_token_accuracy": 0.7963423719406127,
+      "eval_physics_num_tokens": 33075822.0,
+      "eval_physics_runtime": 42.62,
+      "eval_physics_samples_per_second": 11.732,
+      "eval_physics_steps_per_second": 2.933,
+      "step": 1000
+    },
+    {
+      "entropy": 0.7457848465070128,
+      "epoch": 1.616,
+      "grad_norm": 24.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 11.8937,
+      "mean_token_accuracy": 0.8049711644649505,
+      "num_tokens": 33411850.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.7540575532242656,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 23.375,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 12.1774,
+      "mean_token_accuracy": 0.8019963789731264,
+      "num_tokens": 33739456.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.7294436307623983,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 22.875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 11.7158,
+      "mean_token_accuracy": 0.8073111124336719,
+      "num_tokens": 34072022.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.7509479133412242,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 26.0,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 12.0874,
+      "mean_token_accuracy": 0.804041426628828,
+      "num_tokens": 34398959.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.7489136515185237,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 22.625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 11.965,
+      "mean_token_accuracy": 0.8045720625668764,
+      "num_tokens": 34732284.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.7499239794909954,
+      "epoch": 1.696,
+      "grad_norm": 31.625,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 12.076,
+      "mean_token_accuracy": 0.8027914315462112,
+      "num_tokens": 35053712.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.7558027764782309,
+      "epoch": 1.712,
+      "grad_norm": 23.875,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 12.147,
+      "mean_token_accuracy": 0.8005938410758973,
+      "num_tokens": 35393088.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.7422181311994791,
+      "epoch": 1.728,
+      "grad_norm": 25.875,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 11.8962,
+      "mean_token_accuracy": 0.8054456721991301,
+      "num_tokens": 35729007.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.7470023760572075,
+      "epoch": 1.744,
+      "grad_norm": 24.375,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 11.9451,
+      "mean_token_accuracy": 0.8041820522397757,
+      "num_tokens": 36072489.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.736881697922945,
+      "epoch": 1.76,
+      "grad_norm": 25.0,
+      "learning_rate": 1.978e-05,
+      "loss": 11.8338,
+      "mean_token_accuracy": 0.8047639206051826,
+      "num_tokens": 36398285.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6262362420558929,
+      "eval_math_loss": 0.7465543746948242,
+      "eval_math_mean_token_accuracy": 0.805053071975708,
+      "eval_math_num_tokens": 36398285.0,
+      "eval_math_runtime": 37.6479,
+      "eval_math_samples_per_second": 13.281,
+      "eval_math_steps_per_second": 3.32,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.7200349433422089,
+      "eval_physics_loss": 0.7697470188140869,
+      "eval_physics_mean_token_accuracy": 0.8005799927711487,
+      "eval_physics_num_tokens": 36398285.0,
+      "eval_physics_runtime": 42.4278,
+      "eval_physics_samples_per_second": 11.785,
+      "eval_physics_steps_per_second": 2.946,
+      "step": 1100
+    },
+    {
+      "entropy": 0.7551133055239916,
+      "epoch": 1.776,
+      "grad_norm": 22.125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 12.2538,
+      "mean_token_accuracy": 0.7994971070438623,
+      "num_tokens": 36730479.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.7345263505354523,
+      "epoch": 1.792,
+      "grad_norm": 26.5,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 11.7729,
+      "mean_token_accuracy": 0.8065289612859488,
+      "num_tokens": 37057425.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.7239221159368754,
+      "epoch": 1.808,
+      "grad_norm": 20.875,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 11.5735,
+      "mean_token_accuracy": 0.8099185042083263,
+      "num_tokens": 37395449.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.7373615993186832,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 25.625,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 11.8409,
+      "mean_token_accuracy": 0.8052641868591308,
+      "num_tokens": 37729535.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.7255491560325027,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 22.125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 11.71,
+      "mean_token_accuracy": 0.8071777541190386,
+      "num_tokens": 38057839.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.7327572701498866,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 21.375,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 11.7512,
+      "mean_token_accuracy": 0.8067308221012354,
+      "num_tokens": 38398033.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.7250073075294494,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 23.0,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 11.7062,
+      "mean_token_accuracy": 0.804894783347845,
+      "num_tokens": 38729203.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.7499742954969406,
+      "epoch": 1.888,
+      "grad_norm": 24.0,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 12.0141,
+      "mean_token_accuracy": 0.8026742864400148,
+      "num_tokens": 39058791.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.7552636509761215,
+      "epoch": 1.904,
+      "grad_norm": 24.125,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 12.2023,
+      "mean_token_accuracy": 0.801565033942461,
+      "num_tokens": 39386170.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.7536638176068664,
+      "epoch": 1.92,
+      "grad_norm": 22.5,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 12.0831,
+      "mean_token_accuracy": 0.802052004635334,
+      "num_tokens": 39710381.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6647274706363678,
+      "eval_math_loss": 0.7477448582649231,
+      "eval_math_mean_token_accuracy": 0.8044252181053162,
+      "eval_math_num_tokens": 39710381.0,
+      "eval_math_runtime": 37.3147,
+      "eval_math_samples_per_second": 13.4,
+      "eval_math_steps_per_second": 3.35,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.7363031375408172,
+      "eval_physics_loss": 0.7557728290557861,
+      "eval_physics_mean_token_accuracy": 0.8023755269050599,
+      "eval_physics_num_tokens": 39710381.0,
+      "eval_physics_runtime": 42.3841,
+      "eval_physics_samples_per_second": 11.797,
+      "eval_physics_steps_per_second": 2.949,
+      "step": 1200
+    },
+    {
+      "entropy": 0.7332505928352475,
+      "epoch": 1.936,
+      "grad_norm": 23.625,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 11.7747,
+      "mean_token_accuracy": 0.8057302571833134,
+      "num_tokens": 40043354.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6865954516455531,
+      "epoch": 1.952,
+      "grad_norm": 22.875,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 11.0944,
+      "mean_token_accuracy": 0.8145869821310043,
+      "num_tokens": 40375790.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.725792682915926,
+      "epoch": 1.968,
+      "grad_norm": 23.5,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 11.6981,
+      "mean_token_accuracy": 0.807374369353056,
+      "num_tokens": 40705451.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.7336347706615924,
+      "epoch": 1.984,
+      "grad_norm": 22.875,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 11.7816,
+      "mean_token_accuracy": 0.8049177132546902,
+      "num_tokens": 41040313.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.7291126515716314,
+      "epoch": 2.0,
+      "grad_norm": 19.625,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 11.6571,
+      "mean_token_accuracy": 0.8080590449273586,
+      "num_tokens": 41373608.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.6355575416237116,
+      "epoch": 2.016,
+      "grad_norm": 19.875,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 10.132,
+      "mean_token_accuracy": 0.8247552629560232,
+      "num_tokens": 41708226.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.6331347430124878,
+      "epoch": 2.032,
+      "grad_norm": 25.375,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 10.215,
+      "mean_token_accuracy": 0.824965550750494,
+      "num_tokens": 42045433.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.6594979166984558,
+      "epoch": 2.048,
+      "grad_norm": 25.5,
+      "learning_rate": 1.938e-05,
+      "loss": 10.5643,
+      "mean_token_accuracy": 0.8191797107458114,
+      "num_tokens": 42372694.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.6339968785643577,
+      "epoch": 2.064,
+      "grad_norm": 22.25,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 10.1287,
+      "mean_token_accuracy": 0.8256273221224546,
+      "num_tokens": 42710240.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.6453937565907836,
+      "epoch": 2.08,
+      "grad_norm": 22.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 10.3686,
+      "mean_token_accuracy": 0.8218923207372427,
+      "num_tokens": 43037844.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5932842814922332,
+      "eval_math_loss": 0.7640953660011292,
+      "eval_math_mean_token_accuracy": 0.8026974692344665,
+      "eval_math_num_tokens": 43037844.0,
+      "eval_math_runtime": 37.2894,
+      "eval_math_samples_per_second": 13.409,
+      "eval_math_steps_per_second": 3.352,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.6534187440872192,
+      "eval_physics_loss": 0.7533365488052368,
+      "eval_physics_mean_token_accuracy": 0.8034858341217042,
+      "eval_physics_num_tokens": 43037844.0,
+      "eval_physics_runtime": 42.3739,
+      "eval_physics_samples_per_second": 11.8,
+      "eval_physics_steps_per_second": 2.95,
+      "step": 1300
+    },
+    {
+      "entropy": 0.6423438133671879,
+      "epoch": 2.096,
+      "grad_norm": 21.625,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 10.2864,
+      "mean_token_accuracy": 0.8230229053646326,
+      "num_tokens": 43368675.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.628149182535708,
+      "epoch": 2.112,
+      "grad_norm": 24.0,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 10.1059,
+      "mean_token_accuracy": 0.8258296135812998,
+      "num_tokens": 43699544.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.6355112513527275,
+      "epoch": 2.128,
+      "grad_norm": 22.5,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 10.1878,
+      "mean_token_accuracy": 0.8232955761253834,
+      "num_tokens": 44012632.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.6359908759593964,
+      "epoch": 2.144,
+      "grad_norm": 21.0,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 10.1497,
+      "mean_token_accuracy": 0.8242303643375635,
+      "num_tokens": 44349732.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.6610904721543193,
+      "epoch": 2.16,
+      "grad_norm": 21.875,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 10.62,
+      "mean_token_accuracy": 0.8177395671606064,
+      "num_tokens": 44680410.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.6307835169136524,
+      "epoch": 2.176,
+      "grad_norm": 21.0,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 10.1462,
+      "mean_token_accuracy": 0.8254107493907213,
+      "num_tokens": 45017922.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.6242643415927887,
+      "epoch": 2.192,
+      "grad_norm": 22.125,
+      "learning_rate": 1.918e-05,
+      "loss": 10.038,
+      "mean_token_accuracy": 0.8260709717869759,
+      "num_tokens": 45348179.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.6235629940405488,
+      "epoch": 2.208,
+      "grad_norm": 23.5,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 9.993,
+      "mean_token_accuracy": 0.8266679599881173,
+      "num_tokens": 45678397.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.6313284657895565,
+      "epoch": 2.224,
+      "grad_norm": 20.625,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 10.1072,
+      "mean_token_accuracy": 0.8250856719911098,
+      "num_tokens": 46010052.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.6290372915565967,
+      "epoch": 2.24,
+      "grad_norm": 22.125,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 10.0844,
+      "mean_token_accuracy": 0.8260512594133615,
+      "num_tokens": 46347116.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5953125247955322,
+      "eval_math_loss": 0.7714977860450745,
+      "eval_math_mean_token_accuracy": 0.802004508972168,
+      "eval_math_num_tokens": 46347116.0,
+      "eval_math_runtime": 37.2775,
+      "eval_math_samples_per_second": 13.413,
+      "eval_math_steps_per_second": 3.353,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.6626617999076844,
+      "eval_physics_loss": 0.7471906542778015,
+      "eval_physics_mean_token_accuracy": 0.8044270224571228,
+      "eval_physics_num_tokens": 46347116.0,
+      "eval_physics_runtime": 42.3545,
+      "eval_physics_samples_per_second": 11.805,
+      "eval_physics_steps_per_second": 2.951,
+      "step": 1400
+    },
+    {
+      "entropy": 0.6532834043726325,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 23.25,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 10.5479,
+      "mean_token_accuracy": 0.8197706062346697,
+      "num_tokens": 46673074.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.6263921223580837,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 22.875,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 10.0289,
+      "mean_token_accuracy": 0.8267724249511957,
+      "num_tokens": 47009386.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.6413616226986051,
+      "epoch": 2.288,
+      "grad_norm": 24.625,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 10.3579,
+      "mean_token_accuracy": 0.8215750485658646,
+      "num_tokens": 47344231.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.646451156027615,
+      "epoch": 2.304,
+      "grad_norm": 24.625,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 10.3113,
+      "mean_token_accuracy": 0.8226810522377491,
+      "num_tokens": 47658392.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.633142382092774,
+      "epoch": 2.32,
+      "grad_norm": 20.375,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 10.1725,
+      "mean_token_accuracy": 0.8242883045226336,
+      "num_tokens": 47995015.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.6136388331651688,
+      "epoch": 2.336,
+      "grad_norm": 21.125,
+      "learning_rate": 1.898e-05,
+      "loss": 9.8552,
+      "mean_token_accuracy": 0.8278801072388887,
+      "num_tokens": 48336708.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.6158075928688049,
+      "epoch": 2.352,
+      "grad_norm": 21.75,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 9.8776,
+      "mean_token_accuracy": 0.8277330327779054,
+      "num_tokens": 48665311.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.636424052901566,
+      "epoch": 2.368,
+      "grad_norm": 22.0,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 10.1874,
+      "mean_token_accuracy": 0.8247745476663113,
+      "num_tokens": 49005334.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.6439365316182375,
+      "epoch": 2.384,
+      "grad_norm": 24.375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 10.3152,
+      "mean_token_accuracy": 0.8224024541676045,
+      "num_tokens": 49331936.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.6338154060766101,
+      "epoch": 2.4,
+      "grad_norm": 24.375,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 10.2021,
+      "mean_token_accuracy": 0.8241530895233155,
+      "num_tokens": 49671048.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5721300525665283,
+      "eval_math_loss": 0.7793473601341248,
+      "eval_math_mean_token_accuracy": 0.8018091678619385,
+      "eval_math_num_tokens": 49671048.0,
+      "eval_math_runtime": 37.2772,
+      "eval_math_samples_per_second": 13.413,
+      "eval_math_steps_per_second": 3.353,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.6450748410224915,
+      "eval_physics_loss": 0.7388309240341187,
+      "eval_physics_mean_token_accuracy": 0.8056827611923217,
+      "eval_physics_num_tokens": 49671048.0,
+      "eval_physics_runtime": 42.1925,
+      "eval_physics_samples_per_second": 11.85,
+      "eval_physics_steps_per_second": 2.963,
+      "step": 1500
+    },
+    {
+      "entropy": 0.6274364275857807,
+      "epoch": 2.416,
+      "grad_norm": 21.125,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 10.0721,
+      "mean_token_accuracy": 0.8257736101746559,
+      "num_tokens": 50018182.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.634677354246378,
+      "epoch": 2.432,
+      "grad_norm": 25.5,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 10.2384,
+      "mean_token_accuracy": 0.825276605784893,
+      "num_tokens": 50341187.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.6495194172486662,
+      "epoch": 2.448,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 10.4403,
+      "mean_token_accuracy": 0.8209662776440382,
+      "num_tokens": 50667019.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.6421155670657754,
+      "epoch": 2.464,
+      "grad_norm": 23.375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 10.265,
+      "mean_token_accuracy": 0.8245257001370192,
+      "num_tokens": 50991113.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.636642519198358,
+      "epoch": 2.48,
+      "grad_norm": 25.0,
+      "learning_rate": 1.878e-05,
+      "loss": 10.2389,
+      "mean_token_accuracy": 0.8238910645246506,
+      "num_tokens": 51319624.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.6449420966207982,
+      "epoch": 2.496,
+      "grad_norm": 21.75,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 10.293,
+      "mean_token_accuracy": 0.8226108986884355,
+      "num_tokens": 51654633.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.6190562567673623,
+      "epoch": 2.512,
+      "grad_norm": 22.75,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 9.9581,
+      "mean_token_accuracy": 0.8274749897420406,
+      "num_tokens": 51980051.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.6211743861436844,
+      "epoch": 2.528,
+      "grad_norm": 20.25,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 9.9989,
+      "mean_token_accuracy": 0.8274611223489046,
+      "num_tokens": 52316109.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.6385852478444576,
+      "epoch": 2.544,
+      "grad_norm": 24.5,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 10.2452,
+      "mean_token_accuracy": 0.8230951458215714,
+      "num_tokens": 52634528.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.6303336288779974,
+      "epoch": 2.56,
+      "grad_norm": 21.75,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 10.0652,
+      "mean_token_accuracy": 0.8256664223968982,
+      "num_tokens": 52964035.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5893340814113617,
+      "eval_math_loss": 0.777084231376648,
+      "eval_math_mean_token_accuracy": 0.8013126358985901,
+      "eval_math_num_tokens": 52964035.0,
+      "eval_math_runtime": 37.2677,
+      "eval_math_samples_per_second": 13.416,
+      "eval_math_steps_per_second": 3.354,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.6476155369281769,
+      "eval_physics_loss": 0.7298086881637573,
+      "eval_physics_mean_token_accuracy": 0.8077000246047974,
+      "eval_physics_num_tokens": 52964035.0,
+      "eval_physics_runtime": 42.3651,
+      "eval_physics_samples_per_second": 11.802,
+      "eval_physics_steps_per_second": 2.951,
+      "step": 1600
+    },
+    {
+      "entropy": 0.6202353063970805,
+      "epoch": 2.576,
+      "grad_norm": 24.0,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 9.9761,
+      "mean_token_accuracy": 0.8276599138975144,
+      "num_tokens": 53293345.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.6307751543819904,
+      "epoch": 2.592,
+      "grad_norm": 19.625,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 10.1124,
+      "mean_token_accuracy": 0.8253108464181423,
+      "num_tokens": 53620896.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.6150356434285641,
+      "epoch": 2.608,
+      "grad_norm": 19.875,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 9.9156,
+      "mean_token_accuracy": 0.829179810732603,
+      "num_tokens": 53950989.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.6313650351017713,
+      "epoch": 2.624,
+      "grad_norm": 20.375,
+      "learning_rate": 1.858e-05,
+      "loss": 10.121,
+      "mean_token_accuracy": 0.8261329311877489,
+      "num_tokens": 54282732.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.6309516685083508,
+      "epoch": 2.64,
+      "grad_norm": 22.625,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 10.129,
+      "mean_token_accuracy": 0.8251869007945061,
+      "num_tokens": 54609748.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.6598343569785357,
+      "epoch": 2.656,
+      "grad_norm": 23.25,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 10.5853,
+      "mean_token_accuracy": 0.8193065494298934,
+      "num_tokens": 54953583.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.6347682692110539,
+      "epoch": 2.672,
+      "grad_norm": 22.625,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 10.2169,
+      "mean_token_accuracy": 0.8238181065768003,
+      "num_tokens": 55281945.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.6016866784542799,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 22.5,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 9.6888,
+      "mean_token_accuracy": 0.8314425975084305,
+      "num_tokens": 55612391.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.6381715141236782,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 20.75,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 10.1559,
+      "mean_token_accuracy": 0.8256521478295327,
+      "num_tokens": 55949791.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.6189411485567688,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 24.875,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 9.9943,
+      "mean_token_accuracy": 0.8266114149242639,
+      "num_tokens": 56282204.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.5856606528759003,
+      "eval_math_loss": 0.775954008102417,
+      "eval_math_mean_token_accuracy": 0.8015523190498353,
+      "eval_math_num_tokens": 56282204.0,
+      "eval_math_runtime": 37.3399,
+      "eval_math_samples_per_second": 13.391,
+      "eval_math_steps_per_second": 3.348,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.6469226105213165,
+      "eval_physics_loss": 0.7201294302940369,
+      "eval_physics_mean_token_accuracy": 0.8096040639877319,
+      "eval_physics_num_tokens": 56282204.0,
+      "eval_physics_runtime": 42.42,
+      "eval_physics_samples_per_second": 11.787,
+      "eval_physics_steps_per_second": 2.947,
+      "step": 1700
+    },
+    {
+      "entropy": 0.6472665786743164,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 22.25,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 10.3988,
+      "mean_token_accuracy": 0.8214352931827307,
+      "num_tokens": 56605478.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.6382543213665486,
+      "epoch": 2.752,
+      "grad_norm": 21.625,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 10.2692,
+      "mean_token_accuracy": 0.8241269003599883,
+      "num_tokens": 56935874.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.6121929224580527,
+      "epoch": 2.768,
+      "grad_norm": 21.875,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 9.8635,
+      "mean_token_accuracy": 0.8293887037783861,
+      "num_tokens": 57270552.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.631489540822804,
+      "epoch": 2.784,
+      "grad_norm": 23.75,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 10.1392,
+      "mean_token_accuracy": 0.8240107577294111,
+      "num_tokens": 57602968.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.6655854699201882,
+      "epoch": 2.8,
+      "grad_norm": 23.125,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 10.6735,
+      "mean_token_accuracy": 0.8176718927919865,
+      "num_tokens": 57925886.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.6081212440505623,
+      "epoch": 2.816,
+      "grad_norm": 20.125,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 9.6837,
+      "mean_token_accuracy": 0.8310413088649511,
+      "num_tokens": 58262960.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.628994950838387,
+      "epoch": 2.832,
+      "grad_norm": 22.875,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 10.2053,
+      "mean_token_accuracy": 0.8234469190239906,
+      "num_tokens": 58586324.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.6332033973187208,
+      "epoch": 2.848,
+      "grad_norm": 21.875,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 10.0982,
+      "mean_token_accuracy": 0.8255615685135126,
+      "num_tokens": 58912949.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.6201504627242684,
+      "epoch": 2.864,
+      "grad_norm": 19.625,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 9.9752,
+      "mean_token_accuracy": 0.8284646350890398,
+      "num_tokens": 59252494.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.6244674265384674,
+      "epoch": 2.88,
+      "grad_norm": 23.375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 10.0215,
+      "mean_token_accuracy": 0.8271605659276247,
+      "num_tokens": 59574951.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5949274203777313,
+      "eval_math_loss": 0.7759649753570557,
+      "eval_math_mean_token_accuracy": 0.801950795173645,
+      "eval_math_num_tokens": 59574951.0,
+      "eval_math_runtime": 37.6407,
+      "eval_math_samples_per_second": 13.284,
+      "eval_math_steps_per_second": 3.321,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.6501314148902894,
+      "eval_physics_loss": 0.7134683728218079,
+      "eval_physics_mean_token_accuracy": 0.8108458194732666,
+      "eval_physics_num_tokens": 59574951.0,
+      "eval_physics_runtime": 42.4364,
+      "eval_physics_samples_per_second": 11.782,
+      "eval_physics_steps_per_second": 2.946,
+      "step": 1800
+    },
+    {
+      "entropy": 0.6189673518761992,
+      "epoch": 2.896,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 9.9958,
+      "mean_token_accuracy": 0.8274488810449838,
+      "num_tokens": 59903274.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.6149574464187026,
+      "epoch": 2.912,
+      "grad_norm": 21.75,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 9.8638,
+      "mean_token_accuracy": 0.8290023539215327,
+      "num_tokens": 60234203.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.61147581897676,
+      "epoch": 2.928,
+      "grad_norm": 22.375,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 9.778,
+      "mean_token_accuracy": 0.8301452320069075,
+      "num_tokens": 60559123.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.6076565470546484,
+      "epoch": 2.944,
+      "grad_norm": 22.5,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 9.8177,
+      "mean_token_accuracy": 0.8298730112612247,
+      "num_tokens": 60893423.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.6252184329554439,
+      "epoch": 2.96,
+      "grad_norm": 20.125,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 10.0619,
+      "mean_token_accuracy": 0.8257878359407187,
+      "num_tokens": 61228692.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.6309292640537023,
+      "epoch": 2.976,
+      "grad_norm": 22.0,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 10.0273,
+      "mean_token_accuracy": 0.8269333876669407,
+      "num_tokens": 61560852.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.6015587011352181,
+      "epoch": 2.992,
+      "grad_norm": 22.625,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 9.6914,
+      "mean_token_accuracy": 0.8308462183922529,
+      "num_tokens": 61898951.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.554535162076354,
+      "epoch": 3.008,
+      "grad_norm": 32.0,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 8.6037,
+      "mean_token_accuracy": 0.8461754262447357,
+      "num_tokens": 62223875.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.4583619773387909,
+      "epoch": 3.024,
+      "grad_norm": 26.75,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 7.4008,
+      "mean_token_accuracy": 0.8626370068639517,
+      "num_tokens": 62553166.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.45614993553608657,
+      "epoch": 3.04,
+      "grad_norm": 27.625,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 7.303,
+      "mean_token_accuracy": 0.864767599105835,
+      "num_tokens": 62893113.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.4813076241016388,
+      "eval_math_loss": 0.8624137043952942,
+      "eval_math_mean_token_accuracy": 0.7962888913154602,
+      "eval_math_num_tokens": 62893113.0,
+      "eval_math_runtime": 37.3517,
+      "eval_math_samples_per_second": 13.386,
+      "eval_math_steps_per_second": 3.347,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.504680902004242,
+      "eval_physics_loss": 0.7692254781723022,
+      "eval_physics_mean_token_accuracy": 0.806957200050354,
+      "eval_physics_num_tokens": 62893113.0,
+      "eval_physics_runtime": 42.4232,
+      "eval_physics_samples_per_second": 11.786,
+      "eval_physics_steps_per_second": 2.947,
+      "step": 1900
+    },
+    {
+      "entropy": 0.4646745931357145,
+      "epoch": 3.056,
+      "grad_norm": 28.75,
+      "learning_rate": 1.798e-05,
+      "loss": 7.3984,
+      "mean_token_accuracy": 0.8627396196126937,
+      "num_tokens": 63218088.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.4612829519435763,
+      "epoch": 3.072,
+      "grad_norm": 30.25,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 7.3709,
+      "mean_token_accuracy": 0.8628798454999924,
+      "num_tokens": 63551201.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.47167187500745056,
+      "epoch": 3.088,
+      "grad_norm": 29.25,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 7.5434,
+      "mean_token_accuracy": 0.8610360611230135,
+      "num_tokens": 63876362.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.45282970210537316,
+      "epoch": 3.104,
+      "grad_norm": 29.5,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 7.1592,
+      "mean_token_accuracy": 0.8666022688150405,
+      "num_tokens": 64208171.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.44786266181617973,
+      "epoch": 3.12,
+      "grad_norm": 28.75,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 7.1845,
+      "mean_token_accuracy": 0.8669802479445934,
+      "num_tokens": 64544493.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.47433777470141647,
+      "epoch": 3.136,
+      "grad_norm": 28.625,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 7.6129,
+      "mean_token_accuracy": 0.858815074339509,
+      "num_tokens": 64872488.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.45270988158881664,
+      "epoch": 3.152,
+      "grad_norm": 28.25,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 7.2128,
+      "mean_token_accuracy": 0.8667096797376871,
+      "num_tokens": 65201124.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.4651815198361874,
+      "epoch": 3.168,
+      "grad_norm": 27.0,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 7.4436,
+      "mean_token_accuracy": 0.8629633985459805,
+      "num_tokens": 65525868.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.4589499249123037,
+      "epoch": 3.184,
+      "grad_norm": 28.5,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 7.3504,
+      "mean_token_accuracy": 0.8635085135698318,
+      "num_tokens": 65861603.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.46229199739173055,
+      "epoch": 3.2,
+      "grad_norm": 28.0,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 7.4216,
+      "mean_token_accuracy": 0.8631727375090122,
+      "num_tokens": 66193675.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.46265861511230466,
+      "eval_math_loss": 0.8831425309181213,
+      "eval_math_mean_token_accuracy": 0.795429391860962,
+      "eval_math_num_tokens": 66193675.0,
+      "eval_math_runtime": 37.314,
+      "eval_math_samples_per_second": 13.4,
+      "eval_math_steps_per_second": 3.35,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.506291347026825,
+      "eval_physics_loss": 0.77411949634552,
+      "eval_physics_mean_token_accuracy": 0.807035315990448,
+      "eval_physics_num_tokens": 66193675.0,
+      "eval_physics_runtime": 42.2482,
+      "eval_physics_samples_per_second": 11.835,
+      "eval_physics_steps_per_second": 2.959,
+      "step": 2000
+    },
+    {
+      "entropy": 0.46327207051217556,
+      "epoch": 3.216,
+      "grad_norm": 31.375,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 7.3779,
+      "mean_token_accuracy": 0.8627104613929987,
+      "num_tokens": 66522835.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.4881742265075445,
+      "epoch": 3.232,
+      "grad_norm": 28.125,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 7.7779,
+      "mean_token_accuracy": 0.8573839426040649,
+      "num_tokens": 66853693.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.44955099541693927,
+      "epoch": 3.248,
+      "grad_norm": 28.75,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 7.2037,
+      "mean_token_accuracy": 0.8655342176556587,
+      "num_tokens": 67192299.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.4593834072351456,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 31.125,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 7.3846,
+      "mean_token_accuracy": 0.8629851114004851,
+      "num_tokens": 67517068.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.4662083011120558,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 27.625,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 7.4291,
+      "mean_token_accuracy": 0.8624357230961323,
+      "num_tokens": 67841380.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.46831512041389944,
+      "epoch": 3.296,
+      "grad_norm": 28.875,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 7.5193,
+      "mean_token_accuracy": 0.8611114900559187,
+      "num_tokens": 68175287.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.46843817448243497,
+      "epoch": 3.312,
+      "grad_norm": 28.75,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 7.4895,
+      "mean_token_accuracy": 0.8624055199325085,
+      "num_tokens": 68513030.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.4846315659582615,
+      "epoch": 3.328,
+      "grad_norm": 30.5,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 7.7401,
+      "mean_token_accuracy": 0.8579454973340035,
+      "num_tokens": 68839883.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.4725803265348077,
+      "epoch": 3.344,
+      "grad_norm": 32.0,
+      "learning_rate": 1.758e-05,
+      "loss": 7.5094,
+      "mean_token_accuracy": 0.8603448048233986,
+      "num_tokens": 69173450.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.47780570909380915,
+      "epoch": 3.36,
+      "grad_norm": 29.375,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 7.6991,
+      "mean_token_accuracy": 0.8585305999964475,
+      "num_tokens": 69510512.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.47032436895370483,
+      "eval_math_loss": 0.8811061978340149,
+      "eval_math_mean_token_accuracy": 0.7954303793907166,
+      "eval_math_num_tokens": 69510512.0,
+      "eval_math_runtime": 37.313,
+      "eval_math_samples_per_second": 13.4,
+      "eval_math_steps_per_second": 3.35,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.5171641900539398,
+      "eval_physics_loss": 0.7669321298599243,
+      "eval_physics_mean_token_accuracy": 0.8077791213989258,
+      "eval_physics_num_tokens": 69510512.0,
+      "eval_physics_runtime": 42.4208,
+      "eval_physics_samples_per_second": 11.787,
+      "eval_physics_steps_per_second": 2.947,
+      "step": 2100
+    },
+    {
+      "entropy": 0.45618331208825114,
+      "epoch": 3.376,
+      "grad_norm": 27.875,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 7.3065,
+      "mean_token_accuracy": 0.8645048696547747,
+      "num_tokens": 69853992.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.48346296716481446,
+      "epoch": 3.392,
+      "grad_norm": 30.5,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 7.6888,
+      "mean_token_accuracy": 0.856682114303112,
+      "num_tokens": 70178388.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.46400783471763135,
+      "epoch": 3.408,
+      "grad_norm": 29.625,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 7.4566,
+      "mean_token_accuracy": 0.8613572686910629,
+      "num_tokens": 70507903.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.46734738908708096,
+      "epoch": 3.424,
+      "grad_norm": 28.875,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 7.4625,
+      "mean_token_accuracy": 0.8620685461908579,
+      "num_tokens": 70835982.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.4857069244608283,
+      "epoch": 3.44,
+      "grad_norm": 28.0,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 7.8008,
+      "mean_token_accuracy": 0.8561079211533069,
+      "num_tokens": 71172338.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.48103215750306844,
+      "epoch": 3.456,
+      "grad_norm": 28.625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 7.7227,
+      "mean_token_accuracy": 0.8575927283614874,
+      "num_tokens": 71506904.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.48876550355926157,
+      "epoch": 3.472,
+      "grad_norm": 32.0,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 7.8238,
+      "mean_token_accuracy": 0.8549681931734086,
+      "num_tokens": 71830111.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.4620178264565766,
+      "epoch": 3.488,
+      "grad_norm": 28.25,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 7.3903,
+      "mean_token_accuracy": 0.8639769736677408,
+      "num_tokens": 72162307.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.4913556708022952,
+      "epoch": 3.504,
+      "grad_norm": 32.25,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 7.8646,
+      "mean_token_accuracy": 0.8560999035835266,
+      "num_tokens": 72492324.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.478965063393116,
+      "epoch": 3.52,
+      "grad_norm": 29.5,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 7.6389,
+      "mean_token_accuracy": 0.8595894828438759,
+      "num_tokens": 72814868.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.46189505290985106,
+      "eval_math_loss": 0.8891173005104065,
+      "eval_math_mean_token_accuracy": 0.7949126591682434,
+      "eval_math_num_tokens": 72814868.0,
+      "eval_math_runtime": 37.3472,
+      "eval_math_samples_per_second": 13.388,
+      "eval_math_steps_per_second": 3.347,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.5073661119937897,
+      "eval_physics_loss": 0.7688494920730591,
+      "eval_physics_mean_token_accuracy": 0.8081785469055176,
+      "eval_physics_num_tokens": 72814868.0,
+      "eval_physics_runtime": 42.3807,
+      "eval_physics_samples_per_second": 11.798,
+      "eval_physics_steps_per_second": 2.949,
+      "step": 2200
+    },
+    {
+      "entropy": 0.4763089304789901,
+      "epoch": 3.536,
+      "grad_norm": 26.125,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 7.6358,
+      "mean_token_accuracy": 0.8591924477368593,
+      "num_tokens": 73144676.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.47234920123592017,
+      "epoch": 3.552,
+      "grad_norm": 26.375,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 7.5648,
+      "mean_token_accuracy": 0.860359675809741,
+      "num_tokens": 73474150.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.47361249625682833,
+      "epoch": 3.568,
+      "grad_norm": 28.375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 7.5883,
+      "mean_token_accuracy": 0.8600757714360953,
+      "num_tokens": 73814337.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.4998150335624814,
+      "epoch": 3.584,
+      "grad_norm": 28.75,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 8.0356,
+      "mean_token_accuracy": 0.852856045216322,
+      "num_tokens": 74139916.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.4654633553698659,
+      "epoch": 3.6,
+      "grad_norm": 30.625,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 7.4466,
+      "mean_token_accuracy": 0.862101461738348,
+      "num_tokens": 74481410.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.47113845478743316,
+      "epoch": 3.616,
+      "grad_norm": 27.25,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 7.5453,
+      "mean_token_accuracy": 0.8594643916934729,
+      "num_tokens": 74818399.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.4783417483791709,
+      "epoch": 3.632,
+      "grad_norm": 28.0,
+      "learning_rate": 1.718e-05,
+      "loss": 7.6308,
+      "mean_token_accuracy": 0.8589034643024206,
+      "num_tokens": 75148445.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.4815426055341959,
+      "epoch": 3.648,
+      "grad_norm": 28.5,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 7.7338,
+      "mean_token_accuracy": 0.8576061308383942,
+      "num_tokens": 75475448.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.47460295390337703,
+      "epoch": 3.664,
+      "grad_norm": 27.875,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 7.5918,
+      "mean_token_accuracy": 0.8595763813704252,
+      "num_tokens": 75811516.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.48654237780719994,
+      "epoch": 3.68,
+      "grad_norm": 28.375,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 7.8172,
+      "mean_token_accuracy": 0.8558611653745174,
+      "num_tokens": 76131750.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.473254896402359,
+      "eval_math_loss": 0.8871560096740723,
+      "eval_math_mean_token_accuracy": 0.7942537832260131,
+      "eval_math_num_tokens": 76131750.0,
+      "eval_math_runtime": 37.3026,
+      "eval_math_samples_per_second": 13.404,
+      "eval_math_steps_per_second": 3.351,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.5214726667404175,
+      "eval_physics_loss": 0.7622392773628235,
+      "eval_physics_mean_token_accuracy": 0.80874200963974,
+      "eval_physics_num_tokens": 76131750.0,
+      "eval_physics_runtime": 42.3876,
+      "eval_physics_samples_per_second": 11.796,
+      "eval_physics_steps_per_second": 2.949,
+      "step": 2300
+    },
+    {
+      "entropy": 0.5007840525358915,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 30.375,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 7.992,
+      "mean_token_accuracy": 0.8535380810499191,
+      "num_tokens": 76460225.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.4741020784713328,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 29.875,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 7.5979,
+      "mean_token_accuracy": 0.8599839068949222,
+      "num_tokens": 76793614.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.4850239871069789,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 28.0,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 7.8109,
+      "mean_token_accuracy": 0.8563877269625664,
+      "num_tokens": 77121009.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.4851770822890103,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 30.375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 7.7676,
+      "mean_token_accuracy": 0.8573952309787274,
+      "num_tokens": 77446311.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.4895776830613613,
+      "epoch": 3.76,
+      "grad_norm": 28.875,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 7.8327,
+      "mean_token_accuracy": 0.8554267656058073,
+      "num_tokens": 77781488.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.4849891365505755,
+      "epoch": 3.776,
+      "grad_norm": 30.875,
+      "learning_rate": 1.698e-05,
+      "loss": 7.758,
+      "mean_token_accuracy": 0.8568599618971348,
+      "num_tokens": 78108138.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.46994358394294977,
+      "epoch": 3.792,
+      "grad_norm": 28.75,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 7.5618,
+      "mean_token_accuracy": 0.8612664014101028,
+      "num_tokens": 78449177.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.4784676956012845,
+      "epoch": 3.808,
+      "grad_norm": 28.875,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 7.6559,
+      "mean_token_accuracy": 0.8584695469588042,
+      "num_tokens": 78778761.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.49510048162192105,
+      "epoch": 3.824,
+      "grad_norm": 31.375,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 7.8983,
+      "mean_token_accuracy": 0.8549050018191338,
+      "num_tokens": 79102912.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.480623659864068,
+      "epoch": 3.84,
+      "grad_norm": 29.25,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 7.734,
+      "mean_token_accuracy": 0.8581239201128483,
+      "num_tokens": 79436236.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.463162407875061,
+      "eval_math_loss": 0.8948017954826355,
+      "eval_math_mean_token_accuracy": 0.7947532458305359,
+      "eval_math_num_tokens": 79436236.0,
+      "eval_math_runtime": 37.3256,
+      "eval_math_samples_per_second": 13.396,
+      "eval_math_steps_per_second": 3.349,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.5136855947971344,
+      "eval_physics_loss": 0.7588543891906738,
+      "eval_physics_mean_token_accuracy": 0.8093479194641113,
+      "eval_physics_num_tokens": 79436236.0,
+      "eval_physics_runtime": 42.4034,
+      "eval_physics_samples_per_second": 11.791,
+      "eval_physics_steps_per_second": 2.948,
+      "step": 2400
+    },
+    {
+      "entropy": 0.47340967757627367,
+      "epoch": 3.856,
+      "grad_norm": 27.375,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 7.5945,
+      "mean_token_accuracy": 0.8595646426081658,
+      "num_tokens": 79772163.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.48024156149476765,
+      "epoch": 3.872,
+      "grad_norm": 29.375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 7.7006,
+      "mean_token_accuracy": 0.857577583193779,
+      "num_tokens": 80099967.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.48545418698340653,
+      "epoch": 3.888,
+      "grad_norm": 31.125,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 7.8016,
+      "mean_token_accuracy": 0.8571395024657249,
+      "num_tokens": 80430055.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.5100918130017817,
+      "epoch": 3.904,
+      "grad_norm": 35.5,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 8.1543,
+      "mean_token_accuracy": 0.8508139502257108,
+      "num_tokens": 80763923.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.48252020217478275,
+      "epoch": 3.92,
+      "grad_norm": 30.875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 7.7177,
+      "mean_token_accuracy": 0.858184577897191,
+      "num_tokens": 81098568.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.48226326461881397,
+      "epoch": 3.936,
+      "grad_norm": 29.875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 7.7287,
+      "mean_token_accuracy": 0.8575166705995798,
+      "num_tokens": 81423109.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.4868935288861394,
+      "epoch": 3.952,
+      "grad_norm": 29.875,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 7.7548,
+      "mean_token_accuracy": 0.8561869975179434,
+      "num_tokens": 81753214.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.4711942231282592,
+      "epoch": 3.968,
+      "grad_norm": 31.0,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 7.6099,
+      "mean_token_accuracy": 0.8587131906300783,
+      "num_tokens": 82076472.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.46916523650288583,
+      "epoch": 3.984,
+      "grad_norm": 27.625,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 7.5009,
+      "mean_token_accuracy": 0.8613108903169632,
+      "num_tokens": 82418472.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.47474549021571877,
+      "epoch": 4.0,
+      "grad_norm": 31.5,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 7.6216,
+      "mean_token_accuracy": 0.858787452429533,
+      "num_tokens": 82747216.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.47560658168792724,
+      "eval_math_loss": 0.8882602453231812,
+      "eval_math_mean_token_accuracy": 0.7941295528411865,
+      "eval_math_num_tokens": 82747216.0,
+      "eval_math_runtime": 37.5954,
+      "eval_math_samples_per_second": 13.3,
+      "eval_math_steps_per_second": 3.325,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.5194946241378784,
+      "eval_physics_loss": 0.7551596760749817,
+      "eval_physics_mean_token_accuracy": 0.8096559419631958,
+      "eval_physics_num_tokens": 82747216.0,
+      "eval_physics_runtime": 42.2165,
+      "eval_physics_samples_per_second": 11.844,
+      "eval_physics_steps_per_second": 2.961,
+      "step": 2500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.272635579032348e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-2500/training_args.bin ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:c819b77bbc35094ae0bd079366948ce0ef62f469d9ccf51d64318d79c8b67f26
+size 6417