Model save

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,68 @@
+---
+base_model: Qwen/Qwen2.5-1.5B-Instruct
+library_name: transformers
+model_name: Qwen2.5-1.5B-Open-R1-GRPO
+tags:
+- generated_from_trainer
+- trl
+- grpo
+licence: license
+---
+
+# Model Card for Qwen2.5-1.5B-Open-R1-GRPO
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct).
+It has been trained using [TRL](https://github.com/huggingface/trl).
+
+## Quick start
+
+```python
+from transformers import pipeline
+
+question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
+generator = pipeline("text-generation", model="Penghe/Qwen2.5-1.5B-Open-R1-GRPO", device="cuda")
+output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
+print(output["generated_text"])
+```
+
+## Training procedure
+
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/liupenghe1991/huggingface/runs/syvmrocr) 
+
+
+This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
+
+### Framework versions
+
+- TRL: 0.15.0.dev0
+- Transformers: 4.49.0.dev0
+- Pytorch: 2.5.1
+- Datasets: 3.2.0
+- Tokenizers: 0.21.0
+
+## Citations
+
+Cite GRPO as:
+
+```bibtex
+@article{zhihong2024deepseekmath,
+    title        = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
+    author       = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
+    year         = 2024,
+    eprint       = {arXiv:2402.03300},
+}
+
+```
+
+Cite TRL as:
+    
+```bibtex
+@misc{vonwerra2022trl,
+	title        = {{TRL: Transformer Reinforcement Learning}},
+	author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
+	year         = 2020,
+	journal      = {GitHub repository},
+	publisher    = {GitHub},
+	howpublished = {\url{https://github.com/huggingface/trl}}
+}
+```

added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

all_results.json

@@ -0,0 +1,8 @@
+{
+    "total_flos": 0.0,
+    "train_loss": 38.52571455222895,
+    "train_runtime": 51743.6213,
+    "train_samples": 72441,
+    "train_samples_per_second": 1.4,
+    "train_steps_per_second": 0.011
+}

config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "Qwen/Qwen2.5-1.5B-Instruct",
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 21,
+  "model_type": "qwen2",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0.dev0",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.1,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.49.0.dev0"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7a24311e7b0a5d4558c29b5d8da47558152d73d05e206512494fa9bd6cddccd
+size 3087467144

special_tokens_map.json

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

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5eee858c5123a4279c3e1f7b81247343f356ac767940b2692a928ad929543214
+size 11422063

tokenizer_config.json

@@ -0,0 +1,209 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- messages[0]['content'] }}\n    {%- else %}\n        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}\n    {%- endif %}\n    {{- \"\\n\\n# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0]['role'] == 'system' %}\n        {{- '<|im_start|>system\\n' + messages[0]['content'] + '<|im_end|>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>system\\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in messages %}\n    {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) or (message.role == \"assistant\" and not message.tool_calls) %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role }}\n        {%- if message.content %}\n            {{- '\\n' + message.content }}\n        {%- endif %}\n        {%- for tool_call in message.tool_calls %}\n            {%- if tool_call.function is defined %}\n                {%- set tool_call = tool_call.function %}\n            {%- endif %}\n            {{- '\\n<tool_call>\\n{\"name\": \"' }}\n            {{- tool_call.name }}\n            {{- '\", \"arguments\": ' }}\n            {{- tool_call.arguments | tojson }}\n            {{- '}\\n</tool_call>' }}\n        {%- endfor %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "left",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "total_flos": 0.0,
+    "train_loss": 38.52571455222895,
+    "train_runtime": 51743.6213,
+    "train_samples": 72441,
+    "train_samples_per_second": 1.4,
+    "train_steps_per_second": 0.011
+}

trainer_state.json

@@ -0,0 +1,1587 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 100,
+  "global_step": 566,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "completion_length": 390.6765625,
+      "epoch": 0.0088339222614841,
+      "grad_norm": 0.4852478216540892,
+      "kl": 0.0001398921012878418,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 0.0,
+      "reward": 0.615625,
+      "reward_std": 0.2850524147041142,
+      "rewards/accuracy_reward": 0.1515625,
+      "rewards/format_reward": 0.4640625,
+      "step": 5
+    },
+    {
+      "completion_length": 317.87578125,
+      "epoch": 0.0176678445229682,
+      "grad_norm": 1.8804042713820954,
+      "kl": 0.02038722038269043,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 0.0008,
+      "reward": 0.7890625,
+      "reward_std": 0.2585359116084874,
+      "rewards/accuracy_reward": 0.121875,
+      "rewards/format_reward": 0.6671875,
+      "step": 10
+    },
+    {
+      "completion_length": 266.57109375,
+      "epoch": 0.026501766784452298,
+      "grad_norm": 0.4940226477790587,
+      "kl": 0.028921127319335938,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.0012,
+      "reward": 0.85234375,
+      "reward_std": 0.22870484367012978,
+      "rewards/accuracy_reward": 0.0734375,
+      "rewards/format_reward": 0.77890625,
+      "step": 15
+    },
+    {
+      "completion_length": 225.33125,
+      "epoch": 0.0353356890459364,
+      "grad_norm": 1.0484477496661062,
+      "kl": 0.04001922607421875,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 0.0016,
+      "reward": 0.89453125,
+      "reward_std": 0.22870484413579106,
+      "rewards/accuracy_reward": 0.096875,
+      "rewards/format_reward": 0.79765625,
+      "step": 20
+    },
+    {
+      "completion_length": 145.05546875,
+      "epoch": 0.044169611307420496,
+      "grad_norm": 4.764794280687582,
+      "kl": 0.08571319580078125,
+      "learning_rate": 8.771929824561405e-06,
+      "loss": 0.0034,
+      "reward": 0.99609375,
+      "reward_std": 0.16904271068051457,
+      "rewards/accuracy_reward": 0.08125,
+      "rewards/format_reward": 0.91484375,
+      "step": 25
+    },
+    {
+      "completion_length": 185.34609375,
+      "epoch": 0.053003533568904596,
+      "grad_norm": 0.3149380836662221,
+      "kl": 0.040256500244140625,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.0016,
+      "reward": 1.03671875,
+      "reward_std": 0.1447359161451459,
+      "rewards/accuracy_reward": 0.0859375,
+      "rewards/format_reward": 0.95078125,
+      "step": 30
+    },
+    {
+      "completion_length": 228.08125,
+      "epoch": 0.061837455830388695,
+      "grad_norm": 0.2298376401631174,
+      "kl": 0.040625,
+      "learning_rate": 1.2280701754385966e-05,
+      "loss": 0.0016,
+      "reward": 1.04140625,
+      "reward_std": 0.140316498093307,
+      "rewards/accuracy_reward": 0.09375,
+      "rewards/format_reward": 0.94765625,
+      "step": 35
+    },
+    {
+      "completion_length": 180.70234375,
+      "epoch": 0.0706713780918728,
+      "grad_norm": 0.2891998787987488,
+      "kl": 0.05443267822265625,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.0022,
+      "reward": 1.028125,
+      "reward_std": 0.14363106163218617,
+      "rewards/accuracy_reward": 0.08125,
+      "rewards/format_reward": 0.946875,
+      "step": 40
+    },
+    {
+      "completion_length": 199.96484375,
+      "epoch": 0.07950530035335689,
+      "grad_norm": 0.3261101653106404,
+      "kl": 0.06055908203125,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.0024,
+      "reward": 1.03359375,
+      "reward_std": 0.1535747503861785,
+      "rewards/accuracy_reward": 0.09609375,
+      "rewards/format_reward": 0.9375,
+      "step": 45
+    },
+    {
+      "completion_length": 195.6546875,
+      "epoch": 0.08833922261484099,
+      "grad_norm": 0.30278351485149474,
+      "kl": 0.0598968505859375,
+      "learning_rate": 1.754385964912281e-05,
+      "loss": 0.0024,
+      "reward": 1.0640625,
+      "reward_std": 0.15909902211278676,
+      "rewards/accuracy_reward": 0.11484375,
+      "rewards/format_reward": 0.94921875,
+      "step": 50
+    },
+    {
+      "completion_length": 186.865625,
+      "epoch": 0.09717314487632508,
+      "grad_norm": 0.28651839591476297,
+      "kl": 0.0794525146484375,
+      "learning_rate": 1.929824561403509e-05,
+      "loss": 0.0032,
+      "reward": 1.10078125,
+      "reward_std": 0.11821941211819649,
+      "rewards/accuracy_reward": 0.12421875,
+      "rewards/format_reward": 0.9765625,
+      "step": 55
+    },
+    {
+      "completion_length": 162.26328125,
+      "epoch": 0.10600706713780919,
+      "grad_norm": 0.6767776767003258,
+      "kl": 0.09564208984375,
+      "learning_rate": 1.9998285788966027e-05,
+      "loss": 0.0038,
+      "reward": 0.99453125,
+      "reward_std": 0.19113979684188961,
+      "rewards/accuracy_reward": 0.0890625,
+      "rewards/format_reward": 0.90546875,
+      "step": 60
+    },
+    {
+      "completion_length": 202.93203125,
+      "epoch": 0.11484098939929328,
+      "grad_norm": 1179.4543729292784,
+      "kl": 363.3920074462891,
+      "learning_rate": 1.998781218310425e-05,
+      "loss": 14.5041,
+      "reward": 0.903125,
+      "reward_std": 0.2806329967454076,
+      "rewards/accuracy_reward": 0.11328125,
+      "rewards/format_reward": 0.78984375,
+      "step": 65
+    },
+    {
+      "completion_length": 293.38671875,
+      "epoch": 0.12367491166077739,
+      "grad_norm": 0.345762385479784,
+      "kl": 0.248541259765625,
+      "learning_rate": 1.9967827272672407e-05,
+      "loss": 0.0099,
+      "reward": 0.85390625,
+      "reward_std": 0.24859222136437892,
+      "rewards/accuracy_reward": 0.12109375,
+      "rewards/format_reward": 0.7328125,
+      "step": 70
+    },
+    {
+      "completion_length": 111.4828125,
+      "epoch": 0.13250883392226148,
+      "grad_norm": 1.1310414960364419,
+      "kl": 0.187890625,
+      "learning_rate": 1.993835008912268e-05,
+      "loss": 0.0075,
+      "reward": 1.03359375,
+      "reward_std": 0.13810678962618111,
+      "rewards/accuracy_reward": 0.07734375,
+      "rewards/format_reward": 0.95625,
+      "step": 75
+    },
+    {
+      "completion_length": 177.6859375,
+      "epoch": 0.1413427561837456,
+      "grad_norm": 21.313835693015974,
+      "kl": 0.27982177734375,
+      "learning_rate": 1.9899408703314383e-05,
+      "loss": 0.0112,
+      "reward": 1.03515625,
+      "reward_std": 0.12484853835776448,
+      "rewards/accuracy_reward": 0.08984375,
+      "rewards/format_reward": 0.9453125,
+      "step": 80
+    },
+    {
+      "completion_length": 395.63984375,
+      "epoch": 0.1501766784452297,
+      "grad_norm": 0.24256034214749939,
+      "kl": 0.938909912109375,
+      "learning_rate": 1.985104019878233e-05,
+      "loss": 0.0376,
+      "reward": 0.96484375,
+      "reward_std": 0.23754367819055916,
+      "rewards/accuracy_reward": 0.10234375,
+      "rewards/format_reward": 0.8625,
+      "step": 85
+    },
+    {
+      "completion_length": 260.89375,
+      "epoch": 0.15901060070671377,
+      "grad_norm": 0.20016071523180137,
+      "kl": 0.132269287109375,
+      "learning_rate": 1.9793290636422503e-05,
+      "loss": 0.0053,
+      "reward": 1.05859375,
+      "reward_std": 0.1668330023996532,
+      "rewards/accuracy_reward": 0.11640625,
+      "rewards/format_reward": 0.9421875,
+      "step": 90
+    },
+    {
+      "completion_length": 179.8890625,
+      "epoch": 0.16784452296819788,
+      "grad_norm": 0.25156515490686415,
+      "kl": 0.1620361328125,
+      "learning_rate": 1.9726215010628717e-05,
+      "loss": 0.0065,
+      "reward": 1.1421875,
+      "reward_std": 0.13479222664609553,
+      "rewards/accuracy_reward": 0.153125,
+      "rewards/format_reward": 0.9890625,
+      "step": 95
+    },
+    {
+      "completion_length": 197.3421875,
+      "epoch": 0.17667844522968199,
+      "grad_norm": 0.2468861086927419,
+      "kl": 2308.0050537109373,
+      "learning_rate": 1.96498771969219e-05,
+      "loss": 92.1556,
+      "reward": 1.12578125,
+      "reward_std": 0.15357475001364945,
+      "rewards/accuracy_reward": 0.15703125,
+      "rewards/format_reward": 0.96875,
+      "step": 100
+    },
+    {
+      "epoch": 0.17667844522968199,
+      "eval_completion_length": 234.27678571428572,
+      "eval_kl": 0.15457589285714285,
+      "eval_loss": 0.006509294267743826,
+      "eval_reward": 1.188988106591361,
+      "eval_reward_std": 0.1788811758160591,
+      "eval_rewards/accuracy_reward": 0.20238095309053147,
+      "eval_rewards/format_reward": 0.9866071428571429,
+      "eval_runtime": 27.8202,
+      "eval_samples_per_second": 3.559,
+      "eval_steps_per_second": 0.144,
+      "step": 100
+    },
+    {
+      "completion_length": 255.41171875,
+      "epoch": 0.1855123674911661,
+      "grad_norm": 1.7191532060210835,
+      "kl": 106204.82615356446,
+      "learning_rate": 1.9564349891122017e-05,
+      "loss": 4252.0063,
+      "reward": 1.15234375,
+      "reward_std": 0.21323688346892594,
+      "rewards/accuracy_reward": 0.20546875,
+      "rewards/format_reward": 0.946875,
+      "step": 105
+    },
+    {
+      "completion_length": 305.38671875,
+      "epoch": 0.19434628975265017,
+      "grad_norm": 14.1383586771811,
+      "kl": 4.215264892578125,
+      "learning_rate": 1.946971454012051e-05,
+      "loss": 0.1689,
+      "reward": 1.13046875,
+      "reward_std": 0.2795281411148608,
+      "rewards/accuracy_reward": 0.23125,
+      "rewards/format_reward": 0.89921875,
+      "step": 110
+    },
+    {
+      "completion_length": 358.46796875,
+      "epoch": 0.20318021201413428,
+      "grad_norm": 7.705945686711513,
+      "kl": 0.746337890625,
+      "learning_rate": 1.9366061264319112e-05,
+      "loss": 0.0299,
+      "reward": 1.046875,
+      "reward_std": 0.3137786264531314,
+      "rewards/accuracy_reward": 0.2015625,
+      "rewards/format_reward": 0.8453125,
+      "step": 115
+    },
+    {
+      "completion_length": 319.353125,
+      "epoch": 0.21201413427561838,
+      "grad_norm": 0.3374697723436749,
+      "kl": 0.34737548828125,
+      "learning_rate": 1.9253488771809024e-05,
+      "loss": 0.0139,
+      "reward": 1.165625,
+      "reward_std": 0.2828427049331367,
+      "rewards/accuracy_reward": 0.25703125,
+      "rewards/format_reward": 0.90859375,
+      "step": 120
+    },
+    {
+      "completion_length": 271.62265625,
+      "epoch": 0.22084805653710246,
+      "grad_norm": 3.9276115417442674,
+      "kl": 0.31427001953125,
+      "learning_rate": 1.9132104264372065e-05,
+      "loss": 0.0126,
+      "reward": 1.20546875,
+      "reward_std": 0.2264951347373426,
+      "rewards/accuracy_reward": 0.246875,
+      "rewards/format_reward": 0.95859375,
+      "step": 125
+    },
+    {
+      "completion_length": 294.8234375,
+      "epoch": 0.22968197879858657,
+      "grad_norm": 1.7027807029520678,
+      "kl": 0.445849609375,
+      "learning_rate": 1.9002023335393366e-05,
+      "loss": 0.0178,
+      "reward": 1.17421875,
+      "reward_std": 0.23754367902874945,
+      "rewards/accuracy_reward": 0.25078125,
+      "rewards/format_reward": 0.9234375,
+      "step": 130
+    },
+    {
+      "completion_length": 300.08046875,
+      "epoch": 0.23851590106007067,
+      "grad_norm": 0.5632948541759929,
+      "kl": 0.429876708984375,
+      "learning_rate": 1.8863369859782824e-05,
+      "loss": 0.0172,
+      "reward": 1.10078125,
+      "reward_std": 0.24638251382857562,
+      "rewards/accuracy_reward": 0.1890625,
+      "rewards/format_reward": 0.91171875,
+      "step": 135
+    },
+    {
+      "completion_length": 302.4546875,
+      "epoch": 0.24734982332155478,
+      "grad_norm": 17.641244938826887,
+      "kl": 0.55587158203125,
+      "learning_rate": 1.8716275876010135e-05,
+      "loss": 0.0222,
+      "reward": 1.11015625,
+      "reward_std": 0.24196309596300125,
+      "rewards/accuracy_reward": 0.20078125,
+      "rewards/format_reward": 0.909375,
+      "step": 140
+    },
+    {
+      "completion_length": 302.3953125,
+      "epoch": 0.25618374558303886,
+      "grad_norm": 2.250059600074544,
+      "kl": 0.380841064453125,
+      "learning_rate": 1.8560881460365726e-05,
+      "loss": 0.0152,
+      "reward": 1.134375,
+      "reward_std": 0.2585359110496938,
+      "rewards/accuracy_reward": 0.22578125,
+      "rewards/format_reward": 0.90859375,
+      "step": 145
+    },
+    {
+      "completion_length": 326.196875,
+      "epoch": 0.26501766784452296,
+      "grad_norm": 0.47578843736685544,
+      "kl": 0.354583740234375,
+      "learning_rate": 1.8397334593567347e-05,
+      "loss": 0.0142,
+      "reward": 1.1609375,
+      "reward_std": 0.2938912483863533,
+      "rewards/accuracy_reward": 0.265625,
+      "rewards/format_reward": 0.8953125,
+      "step": 150
+    },
+    {
+      "completion_length": 357.03203125,
+      "epoch": 0.27385159010600707,
+      "grad_norm": 0.39094321113316377,
+      "kl": 0.56912841796875,
+      "learning_rate": 1.8225791019839375e-05,
+      "loss": 0.0227,
+      "reward": 1.10078125,
+      "reward_std": 0.3126737717539072,
+      "rewards/accuracy_reward": 0.2359375,
+      "rewards/format_reward": 0.86484375,
+      "step": 155
+    },
+    {
+      "completion_length": 285.365625,
+      "epoch": 0.2826855123674912,
+      "grad_norm": 0.5109380545727407,
+      "kl": 0.448779296875,
+      "learning_rate": 1.8046414098598947e-05,
+      "loss": 0.0179,
+      "reward": 1.12265625,
+      "reward_std": 0.2751087252981961,
+      "rewards/accuracy_reward": 0.22734375,
+      "rewards/format_reward": 0.8953125,
+      "step": 160
+    },
+    {
+      "completion_length": 208.1953125,
+      "epoch": 0.2915194346289753,
+      "grad_norm": 0.7647950522916838,
+      "kl": 1.598681640625,
+      "learning_rate": 1.785937464889027e-05,
+      "loss": 0.0639,
+      "reward": 1.17265625,
+      "reward_std": 0.21323688374832273,
+      "rewards/accuracy_reward": 0.22578125,
+      "rewards/format_reward": 0.946875,
+      "step": 165
+    },
+    {
+      "completion_length": 272.8640625,
+      "epoch": 0.3003533568904594,
+      "grad_norm": 1.2950411415715495,
+      "kl": 0.9397308349609375,
+      "learning_rate": 1.766485078671514e-05,
+      "loss": 0.0376,
+      "reward": 1.08359375,
+      "reward_std": 0.25080193039029836,
+      "rewards/accuracy_reward": 0.19375,
+      "rewards/format_reward": 0.88984375,
+      "step": 170
+    },
+    {
+      "completion_length": 307.93984375,
+      "epoch": 0.30918727915194344,
+      "grad_norm": 1.5917861666741917,
+      "kl": 0.613031005859375,
+      "learning_rate": 1.746302775541467e-05,
+      "loss": 0.0245,
+      "reward": 1.1046875,
+      "reward_std": 0.3115689191967249,
+      "rewards/accuracy_reward": 0.2421875,
+      "rewards/format_reward": 0.8625,
+      "step": 175
+    },
+    {
+      "completion_length": 233.61484375,
+      "epoch": 0.31802120141342755,
+      "grad_norm": 0.39645386165548346,
+      "kl": 0.260223388671875,
+      "learning_rate": 1.7254097749263735e-05,
+      "loss": 0.0104,
+      "reward": 1.1828125,
+      "reward_std": 0.20108348624780775,
+      "rewards/accuracy_reward": 0.24140625,
+      "rewards/format_reward": 0.94140625,
+      "step": 180
+    },
+    {
+      "completion_length": 202.39453125,
+      "epoch": 0.32685512367491165,
+      "grad_norm": 0.33121608670440206,
+      "kl": 1.315484619140625,
+      "learning_rate": 1.703825973044602e-05,
+      "loss": 0.0525,
+      "reward": 1.1984375,
+      "reward_std": 0.181196107622236,
+      "rewards/accuracy_reward": 0.21640625,
+      "rewards/format_reward": 0.98203125,
+      "step": 185
+    },
+    {
+      "completion_length": 270.4625,
+      "epoch": 0.33568904593639576,
+      "grad_norm": 1.7437661744917439,
+      "kl": 0.34991455078125,
+      "learning_rate": 1.681571923958416e-05,
+      "loss": 0.014,
+      "reward": 1.1390625,
+      "reward_std": 0.21213202867656947,
+      "rewards/accuracy_reward": 0.1984375,
+      "rewards/format_reward": 0.940625,
+      "step": 190
+    },
+    {
+      "completion_length": 363.471875,
+      "epoch": 0.34452296819787986,
+      "grad_norm": 1.696436810586237,
+      "kl": 0.6580322265625,
+      "learning_rate": 1.6586688200005193e-05,
+      "loss": 0.0263,
+      "reward": 1.10390625,
+      "reward_std": 0.3038349381648004,
+      "rewards/accuracy_reward": 0.22265625,
+      "rewards/format_reward": 0.88125,
+      "step": 195
+    },
+    {
+      "completion_length": 335.56796875,
+      "epoch": 0.35335689045936397,
+      "grad_norm": 1.1506728460828815,
+      "kl": 0.486822509765625,
+      "learning_rate": 1.6351384715927897e-05,
+      "loss": 0.0195,
+      "reward": 1.13515625,
+      "reward_std": 0.2795281427912414,
+      "rewards/accuracy_reward": 0.25234375,
+      "rewards/format_reward": 0.8828125,
+      "step": 200
+    },
+    {
+      "epoch": 0.35335689045936397,
+      "eval_completion_length": 366.3184552873884,
+      "eval_kl": 0.42898995535714285,
+      "eval_loss": 0.015418421477079391,
+      "eval_reward": 1.0461309552192688,
+      "eval_reward_std": 0.3767027088574001,
+      "eval_rewards/accuracy_reward": 0.20089285714285715,
+      "eval_rewards/format_reward": 0.8452380980764117,
+      "eval_runtime": 42.0295,
+      "eval_samples_per_second": 2.355,
+      "eval_steps_per_second": 0.095,
+      "step": 200
+    },
+    {
+      "completion_length": 330.0671875,
+      "epoch": 0.3621908127208481,
+      "grad_norm": 1.458411327898361,
+      "kl": 0.64564208984375,
+      "learning_rate": 1.611003286476406e-05,
+      "loss": 0.0258,
+      "reward": 1.10859375,
+      "reward_std": 0.3148834802210331,
+      "rewards/accuracy_reward": 0.24140625,
+      "rewards/format_reward": 0.8671875,
+      "step": 205
+    },
+    {
+      "completion_length": 299.528125,
+      "epoch": 0.3710247349823322,
+      "grad_norm": 0.5056086363926139,
+      "kl": 0.61572265625,
+      "learning_rate": 1.5862862483731574e-05,
+      "loss": 0.0246,
+      "reward": 1.1734375,
+      "reward_std": 0.2894718313589692,
+      "rewards/accuracy_reward": 0.2625,
+      "rewards/format_reward": 0.9109375,
+      "step": 210
+    },
+    {
+      "completion_length": 299.7703125,
+      "epoch": 0.37985865724381623,
+      "grad_norm": 0.525532169905154,
+      "kl": 0.350592041015625,
+      "learning_rate": 1.5610108950982494e-05,
+      "loss": 0.014,
+      "reward": 1.12734375,
+      "reward_std": 0.24417280489578844,
+      "rewards/accuracy_reward": 0.21328125,
+      "rewards/format_reward": 0.9140625,
+      "step": 215
+    },
+    {
+      "completion_length": 380.2546875,
+      "epoch": 0.38869257950530034,
+      "grad_norm": 1.1431929583089826,
+      "kl": 24.45216064453125,
+      "learning_rate": 1.535201296145451e-05,
+      "loss": 0.9794,
+      "reward": 1.00703125,
+      "reward_std": 0.30383493890985847,
+      "rewards/accuracy_reward": 0.18046875,
+      "rewards/format_reward": 0.8265625,
+      "step": 220
+    },
+    {
+      "completion_length": 334.7421875,
+      "epoch": 0.39752650176678445,
+      "grad_norm": 0.4471560544674207,
+      "kl": 0.7138427734375,
+      "learning_rate": 1.5088820297659314e-05,
+      "loss": 0.0285,
+      "reward": 1.0984375,
+      "reward_std": 0.24306794982403518,
+      "rewards/accuracy_reward": 0.190625,
+      "rewards/format_reward": 0.9078125,
+      "step": 225
+    },
+    {
+      "completion_length": 296.5125,
+      "epoch": 0.40636042402826855,
+      "grad_norm": 0.6464397657184721,
+      "kl": 0.34703369140625,
+      "learning_rate": 1.4820781595626116e-05,
+      "loss": 0.0139,
+      "reward": 1.15546875,
+      "reward_std": 0.22870484329760074,
+      "rewards/accuracy_reward": 0.23046875,
+      "rewards/format_reward": 0.925,
+      "step": 230
+    },
+    {
+      "completion_length": 348.37578125,
+      "epoch": 0.41519434628975266,
+      "grad_norm": 0.6912675486784617,
+      "kl": 0.5900390625,
+      "learning_rate": 1.4548152106223157e-05,
+      "loss": 0.0236,
+      "reward": 1.075,
+      "reward_std": 0.27842328883707523,
+      "rewards/accuracy_reward": 0.19921875,
+      "rewards/format_reward": 0.87578125,
+      "step": 235
+    },
+    {
+      "completion_length": 279.95859375,
+      "epoch": 0.42402826855123676,
+      "grad_norm": 0.2370174490557891,
+      "kl": 0.32108154296875,
+      "learning_rate": 1.4271191452084598e-05,
+      "loss": 0.0128,
+      "reward": 1.18671875,
+      "reward_std": 0.23091455278918147,
+      "rewards/accuracy_reward": 0.246875,
+      "rewards/format_reward": 0.93984375,
+      "step": 240
+    },
+    {
+      "completion_length": 264.12265625,
+      "epoch": 0.43286219081272087,
+      "grad_norm": 0.2720925122391619,
+      "kl": 0.2543212890625,
+      "learning_rate": 1.3990163380374195e-05,
+      "loss": 0.0102,
+      "reward": 1.13125,
+      "reward_std": 0.20108348615467547,
+      "rewards/accuracy_reward": 0.184375,
+      "rewards/format_reward": 0.946875,
+      "step": 245
+    },
+    {
+      "completion_length": 313.42421875,
+      "epoch": 0.4416961130742049,
+      "grad_norm": 0.353894098469963,
+      "kl": 0.26143798828125,
+      "learning_rate": 1.3705335511621229e-05,
+      "loss": 0.0105,
+      "reward": 1.065625,
+      "reward_std": 0.18561552548781038,
+      "rewards/accuracy_reward": 0.140625,
+      "rewards/format_reward": 0.925,
+      "step": 250
+    },
+    {
+      "completion_length": 330.32734375,
+      "epoch": 0.450530035335689,
+      "grad_norm": 0.2905141185526513,
+      "kl": 0.25023193359375,
+      "learning_rate": 1.3416979084867851e-05,
+      "loss": 0.01,
+      "reward": 1.07265625,
+      "reward_std": 0.20660775769501924,
+      "rewards/accuracy_reward": 0.14609375,
+      "rewards/format_reward": 0.9265625,
+      "step": 255
+    },
+    {
+      "completion_length": 349.93359375,
+      "epoch": 0.45936395759717313,
+      "grad_norm": 1.0004243944782385,
+      "kl": 0.30322265625,
+      "learning_rate": 1.3125368699370567e-05,
+      "loss": 0.0121,
+      "reward": 1.10625,
+      "reward_std": 0.22318057175725697,
+      "rewards/accuracy_reward": 0.18671875,
+      "rewards/format_reward": 0.91953125,
+      "step": 260
+    },
+    {
+      "completion_length": 288.64453125,
+      "epoch": 0.46819787985865724,
+      "grad_norm": 0.6719440192749979,
+      "kl": 0.2525634765625,
+      "learning_rate": 1.2830782053101807e-05,
+      "loss": 0.0101,
+      "reward": 1.13515625,
+      "reward_std": 0.22870484311133624,
+      "rewards/accuracy_reward": 0.1953125,
+      "rewards/format_reward": 0.93984375,
+      "step": 265
+    },
+    {
+      "completion_length": 229.09140625,
+      "epoch": 0.47703180212014135,
+      "grad_norm": 0.7315854476191064,
+      "kl": 0.3944580078125,
+      "learning_rate": 1.2533499678300618e-05,
+      "loss": 0.0158,
+      "reward": 1.1046875,
+      "reward_std": 0.2231805724091828,
+      "rewards/accuracy_reward": 0.1953125,
+      "rewards/format_reward": 0.909375,
+      "step": 270
+    },
+    {
+      "completion_length": 235.57734375,
+      "epoch": 0.48586572438162545,
+      "grad_norm": 0.5879524580475533,
+      "kl": 0.549365234375,
+      "learning_rate": 1.223380467432432e-05,
+      "loss": 0.022,
+      "reward": 1.11953125,
+      "reward_std": 0.23312426200136543,
+      "rewards/accuracy_reward": 0.20546875,
+      "rewards/format_reward": 0.9140625,
+      "step": 275
+    },
+    {
+      "completion_length": 265.6359375,
+      "epoch": 0.49469964664310956,
+      "grad_norm": 1.3544751986419301,
+      "kl": 0.46055908203125,
+      "learning_rate": 1.1931982438055506e-05,
+      "loss": 0.0184,
+      "reward": 1.1328125,
+      "reward_std": 0.23422911558300258,
+      "rewards/accuracy_reward": 0.209375,
+      "rewards/format_reward": 0.9234375,
+      "step": 280
+    },
+    {
+      "completion_length": 248.88203125,
+      "epoch": 0.5035335689045937,
+      "grad_norm": 0.2628588765227769,
+      "kl": 0.315643310546875,
+      "learning_rate": 1.1628320392121118e-05,
+      "loss": 0.0126,
+      "reward": 1.18046875,
+      "reward_std": 0.18893008828163146,
+      "rewards/accuracy_reward": 0.22890625,
+      "rewards/format_reward": 0.9515625,
+      "step": 285
+    },
+    {
+      "completion_length": 234.8484375,
+      "epoch": 0.5123674911660777,
+      "grad_norm": 0.4080469891615222,
+      "kl": 0.19666748046875,
+      "learning_rate": 1.1323107711182474e-05,
+      "loss": 0.0079,
+      "reward": 1.1984375,
+      "reward_std": 0.19003494316712022,
+      "rewards/accuracy_reward": 0.22578125,
+      "rewards/format_reward": 0.97265625,
+      "step": 290
+    },
+    {
+      "completion_length": 266.90078125,
+      "epoch": 0.5212014134275619,
+      "grad_norm": 2.4756632872358195,
+      "kl": 0.452166748046875,
+      "learning_rate": 1.1016635046556773e-05,
+      "loss": 0.018,
+      "reward": 1.153125,
+      "reward_std": 0.20108348615467547,
+      "rewards/accuracy_reward": 0.209375,
+      "rewards/format_reward": 0.94375,
+      "step": 295
+    },
+    {
+      "completion_length": 343.42890625,
+      "epoch": 0.5300353356890459,
+      "grad_norm": 0.9813409241888567,
+      "kl": 0.79124755859375,
+      "learning_rate": 1.0709194249432471e-05,
+      "loss": 0.0316,
+      "reward": 1.046875,
+      "reward_std": 0.24748736610636116,
+      "rewards/accuracy_reward": 0.17109375,
+      "rewards/format_reward": 0.87578125,
+      "step": 300
+    },
+    {
+      "epoch": 0.5300353356890459,
+      "eval_completion_length": 324.2782745361328,
+      "eval_kl": 0.7494419642857143,
+      "eval_loss": 0.03394850715994835,
+      "eval_reward": 1.03125,
+      "eval_reward_std": 0.23359777459076472,
+      "eval_rewards/accuracy_reward": 0.13839285714285715,
+      "eval_rewards/format_reward": 0.8928571428571429,
+      "eval_runtime": 45.797,
+      "eval_samples_per_second": 2.162,
+      "eval_steps_per_second": 0.087,
+      "step": 300
+    },
+    {
+      "completion_length": 311.45,
+      "epoch": 0.5388692579505301,
+      "grad_norm": 0.3646006498120837,
+      "kl": 0.431866455078125,
+      "learning_rate": 1.0401078092941972e-05,
+      "loss": 0.0173,
+      "reward": 1.1171875,
+      "reward_std": 0.2231805719435215,
+      "rewards/accuracy_reward": 0.19453125,
+      "rewards/format_reward": 0.92265625,
+      "step": 305
+    },
+    {
+      "completion_length": 319.184375,
+      "epoch": 0.5477031802120141,
+      "grad_norm": 0.47201265926523844,
+      "kl": 0.28321533203125,
+      "learning_rate": 1.0092579993356386e-05,
+      "loss": 0.0113,
+      "reward": 1.15234375,
+      "reward_std": 0.22649513520300388,
+      "rewards/accuracy_reward": 0.21875,
+      "rewards/format_reward": 0.93359375,
+      "step": 310
+    },
+    {
+      "completion_length": 332.465625,
+      "epoch": 0.5565371024734982,
+      "grad_norm": 0.25398784981603234,
+      "kl": 0.299932861328125,
+      "learning_rate": 9.783993730667833e-06,
+      "loss": 0.012,
+      "reward": 1.17265625,
+      "reward_std": 0.2618504726327956,
+      "rewards/accuracy_reward": 0.259375,
+      "rewards/format_reward": 0.91328125,
+      "step": 315
+    },
+    {
+      "completion_length": 285.19140625,
+      "epoch": 0.5653710247349824,
+      "grad_norm": 0.22747963506264485,
+      "kl": 0.150909423828125,
+      "learning_rate": 9.475613168825374e-06,
+      "loss": 0.006,
+      "reward": 1.2125,
+      "reward_std": 0.18119610771536826,
+      "rewards/accuracy_reward": 0.2515625,
+      "rewards/format_reward": 0.9609375,
+      "step": 320
+    },
+    {
+      "completion_length": 285.81796875,
+      "epoch": 0.5742049469964664,
+      "grad_norm": 0.24723365093480615,
+      "kl": 0.171392822265625,
+      "learning_rate": 9.167731975890977e-06,
+      "loss": 0.0069,
+      "reward": 1.2078125,
+      "reward_std": 0.21434173732995987,
+      "rewards/accuracy_reward": 0.25703125,
+      "rewards/format_reward": 0.95078125,
+      "step": 325
+    },
+    {
+      "completion_length": 335.215625,
+      "epoch": 0.5830388692579506,
+      "grad_norm": 1.4763406749728052,
+      "kl": 0.252349853515625,
+      "learning_rate": 8.860643344382057e-06,
+      "loss": 0.0101,
+      "reward": 1.1140625,
+      "reward_std": 0.24969707559794188,
+      "rewards/accuracy_reward": 0.20390625,
+      "rewards/format_reward": 0.91015625,
+      "step": 330
+    },
+    {
+      "completion_length": 307.25625,
+      "epoch": 0.5918727915194346,
+      "grad_norm": 0.3672971332309817,
+      "kl": 0.25814208984375,
+      "learning_rate": 8.554639712066837e-06,
+      "loss": 0.0103,
+      "reward": 1.1125,
+      "reward_std": 0.27621358102187515,
+      "rewards/accuracy_reward": 0.215625,
+      "rewards/format_reward": 0.896875,
+      "step": 335
+    },
+    {
+      "completion_length": 242.44296875,
+      "epoch": 0.6007067137809188,
+      "grad_norm": 0.4368486045003144,
+      "kl": 0.190582275390625,
+      "learning_rate": 8.250012483478478e-06,
+      "loss": 0.0076,
+      "reward": 1.16640625,
+      "reward_std": 0.20439804838970305,
+      "rewards/accuracy_reward": 0.21953125,
+      "rewards/format_reward": 0.946875,
+      "step": 340
+    },
+    {
+      "completion_length": 225.3546875,
+      "epoch": 0.6095406360424028,
+      "grad_norm": 0.42905104827822127,
+      "kl": 0.263653564453125,
+      "learning_rate": 7.947051752413131e-06,
+      "loss": 0.0105,
+      "reward": 1.18125,
+      "reward_std": 0.21876115379855038,
+      "rewards/accuracy_reward": 0.24453125,
+      "rewards/format_reward": 0.93671875,
+      "step": 345
+    },
+    {
+      "completion_length": 230.521875,
+      "epoch": 0.6183745583038869,
+      "grad_norm": 0.3770160036076146,
+      "kl": 0.329302978515625,
+      "learning_rate": 7.646046025676198e-06,
+      "loss": 0.0132,
+      "reward": 1.08203125,
+      "reward_std": 0.26626989198848605,
+      "rewards/accuracy_reward": 0.1859375,
+      "rewards/format_reward": 0.89609375,
+      "step": 350
+    },
+    {
+      "completion_length": 194.5765625,
+      "epoch": 0.627208480565371,
+      "grad_norm": 0.2624186504175127,
+      "kl": 0.2316162109375,
+      "learning_rate": 7.34728194833988e-06,
+      "loss": 0.0093,
+      "reward": 1.15390625,
+      "reward_std": 0.20660775676369666,
+      "rewards/accuracy_reward": 0.209375,
+      "rewards/format_reward": 0.94453125,
+      "step": 355
+    },
+    {
+      "completion_length": 235.79921875,
+      "epoch": 0.6360424028268551,
+      "grad_norm": 2.783327950557256,
+      "kl": 0.3891357421875,
+      "learning_rate": 7.051044030773619e-06,
+      "loss": 0.0156,
+      "reward": 1.14921875,
+      "reward_std": 0.28394756009802224,
+      "rewards/accuracy_reward": 0.23671875,
+      "rewards/format_reward": 0.9125,
+      "step": 360
+    },
+    {
+      "completion_length": 264.353125,
+      "epoch": 0.6448763250883393,
+      "grad_norm": 0.36191769311818617,
+      "kl": 0.31597900390625,
+      "learning_rate": 6.757614377707409e-06,
+      "loss": 0.0126,
+      "reward": 1.15,
+      "reward_std": 0.22980969855561853,
+      "rewards/accuracy_reward": 0.23671875,
+      "rewards/format_reward": 0.91328125,
+      "step": 365
+    },
+    {
+      "completion_length": 259.45,
+      "epoch": 0.6537102473498233,
+      "grad_norm": 0.39476079322309915,
+      "kl": 0.17608642578125,
+      "learning_rate": 6.467272419585984e-06,
+      "loss": 0.007,
+      "reward": 1.24296875,
+      "reward_std": 0.20660775816068053,
+      "rewards/accuracy_reward": 0.28046875,
+      "rewards/format_reward": 0.9625,
+      "step": 370
+    },
+    {
+      "completion_length": 291.5421875,
+      "epoch": 0.6625441696113075,
+      "grad_norm": 0.40239608127232374,
+      "kl": 0.223699951171875,
+      "learning_rate": 6.18029464646968e-06,
+      "loss": 0.009,
+      "reward": 1.19921875,
+      "reward_std": 0.22207571724429725,
+      "rewards/accuracy_reward": 0.24921875,
+      "rewards/format_reward": 0.95,
+      "step": 375
+    },
+    {
+      "completion_length": 329.928125,
+      "epoch": 0.6713780918727915,
+      "grad_norm": 0.8357308043400351,
+      "kl": 0.220965576171875,
+      "learning_rate": 5.896954344735426e-06,
+      "loss": 0.0088,
+      "reward": 1.15703125,
+      "reward_std": 0.24417280619964005,
+      "rewards/accuracy_reward": 0.23203125,
+      "rewards/format_reward": 0.925,
+      "step": 380
+    },
+    {
+      "completion_length": 313.63984375,
+      "epoch": 0.6802120141342756,
+      "grad_norm": 0.3999919576901151,
+      "kl": 0.2136962890625,
+      "learning_rate": 5.617521336828556e-06,
+      "loss": 0.0086,
+      "reward": 1.21171875,
+      "reward_std": 0.23091455288231372,
+      "rewards/accuracy_reward": 0.26875,
+      "rewards/format_reward": 0.94296875,
+      "step": 385
+    },
+    {
+      "completion_length": 343.1890625,
+      "epoch": 0.6890459363957597,
+      "grad_norm": 0.579022803109722,
+      "kl": 0.20040283203125,
+      "learning_rate": 5.342261724313292e-06,
+      "loss": 0.008,
+      "reward": 1.1390625,
+      "reward_std": 0.2762135806493461,
+      "rewards/accuracy_reward": 0.24296875,
+      "rewards/format_reward": 0.89609375,
+      "step": 390
+    },
+    {
+      "completion_length": 325.75234375,
+      "epoch": 0.6978798586572438,
+      "grad_norm": 0.3145860906634086,
+      "kl": 0.181256103515625,
+      "learning_rate": 5.0714376344666095e-06,
+      "loss": 0.0072,
+      "reward": 1.1296875,
+      "reward_std": 0.27400387059897185,
+      "rewards/accuracy_reward": 0.2203125,
+      "rewards/format_reward": 0.909375,
+      "step": 395
+    },
+    {
+      "completion_length": 280.128125,
+      "epoch": 0.7067137809187279,
+      "grad_norm": 0.6989923168355932,
+      "kl": 0.1529296875,
+      "learning_rate": 4.8053069706567555e-06,
+      "loss": 0.0061,
+      "reward": 1.175,
+      "reward_std": 0.20550290299579502,
+      "rewards/accuracy_reward": 0.228125,
+      "rewards/format_reward": 0.946875,
+      "step": 400
+    },
+    {
+      "epoch": 0.7067137809187279,
+      "eval_completion_length": 253.9434531075614,
+      "eval_kl": 0.14620535714285715,
+      "eval_loss": 0.006245024502277374,
+      "eval_reward": 1.1904761961528234,
+      "eval_reward_std": 0.1767766922712326,
+      "eval_rewards/accuracy_reward": 0.23511904903820582,
+      "eval_rewards/format_reward": 0.9553571428571429,
+      "eval_runtime": 44.1368,
+      "eval_samples_per_second": 2.243,
+      "eval_steps_per_second": 0.091,
+      "step": 400
+    },
+    {
+      "completion_length": 282.00234375,
+      "epoch": 0.715547703180212,
+      "grad_norm": 0.37937913622208536,
+      "kl": 0.169927978515625,
+      "learning_rate": 4.5441231667441724e-06,
+      "loss": 0.0068,
+      "reward": 1.125,
+      "reward_std": 0.23864853335544467,
+      "rewards/accuracy_reward": 0.203125,
+      "rewards/format_reward": 0.921875,
+      "step": 405
+    },
+    {
+      "completion_length": 324.4953125,
+      "epoch": 0.7243816254416962,
+      "grad_norm": 0.5286138921912154,
+      "kl": 0.212841796875,
+      "learning_rate": 4.288134945738684e-06,
+      "loss": 0.0085,
+      "reward": 1.10546875,
+      "reward_std": 0.2795281424187124,
+      "rewards/accuracy_reward": 0.2140625,
+      "rewards/format_reward": 0.89140625,
+      "step": 410
+    },
+    {
+      "completion_length": 301.0734375,
+      "epoch": 0.7332155477031802,
+      "grad_norm": 0.39288088936432175,
+      "kl": 0.163116455078125,
+      "learning_rate": 4.037586082942805e-06,
+      "loss": 0.0065,
+      "reward": 1.17578125,
+      "reward_std": 0.24196309484541417,
+      "rewards/accuracy_reward": 0.25234375,
+      "rewards/format_reward": 0.9234375,
+      "step": 415
+    },
+    {
+      "completion_length": 310.3390625,
+      "epoch": 0.7420494699646644,
+      "grad_norm": 0.1939857255562333,
+      "kl": 0.1360107421875,
+      "learning_rate": 3.7927151738066693e-06,
+      "loss": 0.0054,
+      "reward": 1.196875,
+      "reward_std": 0.19003494326025247,
+      "rewards/accuracy_reward": 0.23984375,
+      "rewards/format_reward": 0.95703125,
+      "step": 420
+    },
+    {
+      "completion_length": 293.96328125,
+      "epoch": 0.7508833922261484,
+      "grad_norm": 0.47246106849789127,
+      "kl": 0.140972900390625,
+      "learning_rate": 3.553755406715724e-06,
+      "loss": 0.0056,
+      "reward": 1.20078125,
+      "reward_std": 0.20660775722935795,
+      "rewards/accuracy_reward": 0.24453125,
+      "rewards/format_reward": 0.95625,
+      "step": 425
+    },
+    {
+      "completion_length": 306.07109375,
+      "epoch": 0.7597173144876325,
+      "grad_norm": 0.4661050396886567,
+      "kl": 0.173931884765625,
+      "learning_rate": 3.320934340927513e-06,
+      "loss": 0.007,
+      "reward": 1.15859375,
+      "reward_std": 0.23312426200136543,
+      "rewards/accuracy_reward": 0.22265625,
+      "rewards/format_reward": 0.9359375,
+      "step": 430
+    },
+    {
+      "completion_length": 350.16875,
+      "epoch": 0.7685512367491166,
+      "grad_norm": 0.2742809206837861,
+      "kl": 0.21610107421875,
+      "learning_rate": 3.094473689869002e-06,
+      "loss": 0.0086,
+      "reward": 1.11171875,
+      "reward_std": 0.25301163820549843,
+      "rewards/accuracy_reward": 0.2140625,
+      "rewards/format_reward": 0.89765625,
+      "step": 435
+    },
+    {
+      "completion_length": 292.36015625,
+      "epoch": 0.7773851590106007,
+      "grad_norm": 0.5350758709986372,
+      "kl": 0.152239990234375,
+      "learning_rate": 2.8745891100008683e-06,
+      "loss": 0.0061,
+      "reward": 1.19609375,
+      "reward_std": 0.22649513492360712,
+      "rewards/accuracy_reward": 0.253125,
+      "rewards/format_reward": 0.94296875,
+      "step": 440
+    },
+    {
+      "completion_length": 312.64296875,
+      "epoch": 0.7862190812720848,
+      "grad_norm": 0.39936217998818846,
+      "kl": 0.15882568359375,
+      "learning_rate": 2.6614899954497797e-06,
+      "loss": 0.0064,
+      "reward": 1.17421875,
+      "reward_std": 0.2552213478833437,
+      "rewards/accuracy_reward": 0.246875,
+      "rewards/format_reward": 0.92734375,
+      "step": 445
+    },
+    {
+      "completion_length": 296.9109375,
+      "epoch": 0.7950530035335689,
+      "grad_norm": 0.5729272885536102,
+      "kl": 0.171441650390625,
+      "learning_rate": 2.455379278604226e-06,
+      "loss": 0.0069,
+      "reward": 1.1671875,
+      "reward_std": 0.26074561988934875,
+      "rewards/accuracy_reward": 0.2390625,
+      "rewards/format_reward": 0.928125,
+      "step": 450
+    },
+    {
+      "completion_length": 278.128125,
+      "epoch": 0.803886925795053,
+      "grad_norm": 0.440915873156984,
+      "kl": 0.192681884765625,
+      "learning_rate": 2.256453236863815e-06,
+      "loss": 0.0077,
+      "reward": 1.1515625,
+      "reward_std": 0.22980969725176692,
+      "rewards/accuracy_reward": 0.2265625,
+      "rewards/format_reward": 0.925,
+      "step": 455
+    },
+    {
+      "completion_length": 252.37578125,
+      "epoch": 0.8127208480565371,
+      "grad_norm": 0.30325471449188185,
+      "kl": 0.165386962890625,
+      "learning_rate": 2.064901305726055e-06,
+      "loss": 0.0066,
+      "reward": 1.1640625,
+      "reward_std": 0.22318057250231504,
+      "rewards/accuracy_reward": 0.22109375,
+      "rewards/format_reward": 0.94296875,
+      "step": 460
+    },
+    {
+      "completion_length": 249.725,
+      "epoch": 0.8215547703180212,
+      "grad_norm": 0.3292435097242347,
+      "kl": 0.300885009765625,
+      "learning_rate": 1.880905898388612e-06,
+      "loss": 0.0121,
+      "reward": 1.159375,
+      "reward_std": 0.2275999895296991,
+      "rewards/accuracy_reward": 0.221875,
+      "rewards/format_reward": 0.9375,
+      "step": 465
+    },
+    {
+      "completion_length": 260.39921875,
+      "epoch": 0.8303886925795053,
+      "grad_norm": 0.33053969954297574,
+      "kl": 0.17269287109375,
+      "learning_rate": 1.7046422320388556e-06,
+      "loss": 0.0069,
+      "reward": 1.13984375,
+      "reward_std": 0.244172805827111,
+      "rewards/accuracy_reward": 0.2125,
+      "rewards/format_reward": 0.92734375,
+      "step": 470
+    },
+    {
+      "completion_length": 276.50546875,
+      "epoch": 0.8392226148409894,
+      "grad_norm": 0.4076018115425754,
+      "kl": 0.18126220703125,
+      "learning_rate": 1.5362781609960853e-06,
+      "loss": 0.0072,
+      "reward": 1.14609375,
+      "reward_std": 0.25964076481759546,
+      "rewards/accuracy_reward": 0.2296875,
+      "rewards/format_reward": 0.91640625,
+      "step": 475
+    },
+    {
+      "completion_length": 293.95859375,
+      "epoch": 0.8480565371024735,
+      "grad_norm": 0.3391066264222914,
+      "kl": 0.220306396484375,
+      "learning_rate": 1.375974016865359e-06,
+      "loss": 0.0088,
+      "reward": 1.14375,
+      "reward_std": 0.2938912480138242,
+      "rewards/accuracy_reward": 0.246875,
+      "rewards/format_reward": 0.896875,
+      "step": 480
+    },
+    {
+      "completion_length": 298.94140625,
+      "epoch": 0.8568904593639576,
+      "grad_norm": 0.2791781526436182,
+      "kl": 0.188665771484375,
+      "learning_rate": 1.2238824558551365e-06,
+      "loss": 0.0076,
+      "reward": 1.121875,
+      "reward_std": 0.289471832010895,
+      "rewards/accuracy_reward": 0.22578125,
+      "rewards/format_reward": 0.89609375,
+      "step": 485
+    },
+    {
+      "completion_length": 285.52734375,
+      "epoch": 0.8657243816254417,
+      "grad_norm": 0.38749404443415436,
+      "kl": 0.19932861328125,
+      "learning_rate": 1.080148313404127e-06,
+      "loss": 0.008,
+      "reward": 1.12265625,
+      "reward_std": 0.2397533869370818,
+      "rewards/accuracy_reward": 0.21875,
+      "rewards/format_reward": 0.90390625,
+      "step": 490
+    },
+    {
+      "completion_length": 294.83203125,
+      "epoch": 0.8745583038869258,
+      "grad_norm": 0.3377327424626226,
+      "kl": 0.173272705078125,
+      "learning_rate": 9.449084662557984e-07,
+      "loss": 0.0069,
+      "reward": 1.11953125,
+      "reward_std": 0.2574310560710728,
+      "rewards/accuracy_reward": 0.21875,
+      "rewards/format_reward": 0.90078125,
+      "step": 495
+    },
+    {
+      "completion_length": 291.9875,
+      "epoch": 0.8833922261484098,
+      "grad_norm": 0.6684225491553946,
+      "kl": 0.201971435546875,
+      "learning_rate": 8.182917021118664e-07,
+      "loss": 0.0081,
+      "reward": 1.121875,
+      "reward_std": 0.26516503673046826,
+      "rewards/accuracy_reward": 0.21953125,
+      "rewards/format_reward": 0.90234375,
+      "step": 500
+    },
+    {
+      "epoch": 0.8833922261484098,
+      "eval_completion_length": 246.29166739327567,
+      "eval_kl": 0.22893415178571427,
+      "eval_loss": 0.009715121239423752,
+      "eval_reward": 1.1130952494485038,
+      "eval_reward_std": 0.26095607025282724,
+      "eval_rewards/accuracy_reward": 0.21279762046677725,
+      "eval_rewards/format_reward": 0.9002976247242519,
+      "eval_runtime": 45.3138,
+      "eval_samples_per_second": 2.185,
+      "eval_steps_per_second": 0.088,
+      "step": 500
+    },
+    {
+      "completion_length": 279.815625,
+      "epoch": 0.892226148409894,
+      "grad_norm": 0.253219550163686,
+      "kl": 0.16181640625,
+      "learning_rate": 7.004185969889188e-07,
+      "loss": 0.0065,
+      "reward": 1.1296875,
+      "reward_std": 0.24306794991716743,
+      "rewards/accuracy_reward": 0.21015625,
+      "rewards/format_reward": 0.91953125,
+      "step": 505
+    },
+    {
+      "completion_length": 261.96171875,
+      "epoch": 0.901060070671378,
+      "grad_norm": 0.31362667762326296,
+      "kl": 0.181817626953125,
+      "learning_rate": 5.914014003949408e-07,
+      "loss": 0.0073,
+      "reward": 1.196875,
+      "reward_std": 0.22759998943656684,
+      "rewards/accuracy_reward": 0.25703125,
+      "rewards/format_reward": 0.93984375,
+      "step": 510
+    },
+    {
+      "completion_length": 255.5484375,
+      "epoch": 0.9098939929328622,
+      "grad_norm": 0.350457592609222,
+      "kl": 0.160760498046875,
+      "learning_rate": 4.913439284351207e-07,
+      "loss": 0.0064,
+      "reward": 1.159375,
+      "reward_std": 0.21876115361228585,
+      "rewards/accuracy_reward": 0.221875,
+      "rewards/format_reward": 0.9375,
+      "step": 515
+    },
+    {
+      "completion_length": 257.99609375,
+      "epoch": 0.9187279151943463,
+      "grad_norm": 0.35463995011806837,
+      "kl": 0.202685546875,
+      "learning_rate": 4.003414649486892e-07,
+      "loss": 0.0081,
+      "reward": 1.18515625,
+      "reward_std": 0.2508019292727113,
+      "rewards/accuracy_reward": 0.2515625,
+      "rewards/format_reward": 0.93359375,
+      "step": 520
+    },
+    {
+      "completion_length": 258.81171875,
+      "epoch": 0.9275618374558304,
+      "grad_norm": 0.40405329273856144,
+      "kl": 0.157171630859375,
+      "learning_rate": 3.184806707709698e-07,
+      "loss": 0.0063,
+      "reward": 1.18125,
+      "reward_std": 0.22097086440771818,
+      "rewards/accuracy_reward": 0.24609375,
+      "rewards/format_reward": 0.93515625,
+      "step": 525
+    },
+    {
+      "completion_length": 275.76796875,
+      "epoch": 0.9363957597173145,
+      "grad_norm": 0.3312261047899013,
+      "kl": 0.1757080078125,
+      "learning_rate": 2.458395012070369e-07,
+      "loss": 0.007,
+      "reward": 1.16875,
+      "reward_std": 0.2452776593156159,
+      "rewards/accuracy_reward": 0.2421875,
+      "rewards/format_reward": 0.9265625,
+      "step": 530
+    },
+    {
+      "completion_length": 262.50546875,
+      "epoch": 0.9452296819787986,
+      "grad_norm": 0.29167819846444787,
+      "kl": 0.15113525390625,
+      "learning_rate": 1.8248713179557788e-07,
+      "loss": 0.006,
+      "reward": 1.175,
+      "reward_std": 0.23422911567613483,
+      "rewards/accuracy_reward": 0.240625,
+      "rewards/format_reward": 0.934375,
+      "step": 535
+    },
+    {
+      "completion_length": 273.60546875,
+      "epoch": 0.9540636042402827,
+      "grad_norm": 0.27251571599773367,
+      "kl": 0.165777587890625,
+      "learning_rate": 1.2848389243363514e-07,
+      "loss": 0.0066,
+      "reward": 1.1484375,
+      "reward_std": 0.25853590965270995,
+      "rewards/accuracy_reward": 0.22421875,
+      "rewards/format_reward": 0.92421875,
+      "step": 540
+    },
+    {
+      "completion_length": 250.08359375,
+      "epoch": 0.9628975265017667,
+      "grad_norm": 0.9804272332040647,
+      "kl": 0.174853515625,
+      "learning_rate": 8.388120992499083e-08,
+      "loss": 0.007,
+      "reward": 1.1546875,
+      "reward_std": 0.22097086338326336,
+      "rewards/accuracy_reward": 0.21953125,
+      "rewards/format_reward": 0.93515625,
+      "step": 545
+    },
+    {
+      "completion_length": 271.278125,
+      "epoch": 0.9717314487632509,
+      "grad_norm": 0.4298853712441631,
+      "kl": 0.19146728515625,
+      "learning_rate": 4.8721559006873473e-08,
+      "loss": 0.0077,
+      "reward": 1.15859375,
+      "reward_std": 0.2287048440426588,
+      "rewards/accuracy_reward": 0.2390625,
+      "rewards/format_reward": 0.91953125,
+      "step": 550
+    },
+    {
+      "completion_length": 270.33515625,
+      "epoch": 0.980565371024735,
+      "grad_norm": 0.38135678110300186,
+      "kl": 0.1751953125,
+      "learning_rate": 2.3038421901651064e-08,
+      "loss": 0.007,
+      "reward": 1.15625,
+      "reward_std": 0.24527765866369008,
+      "rewards/accuracy_reward": 0.22578125,
+      "rewards/format_reward": 0.93046875,
+      "step": 555
+    },
+    {
+      "completion_length": 271.0609375,
+      "epoch": 0.9893992932862191,
+      "grad_norm": 0.3966076226323456,
+      "kl": 0.208819580078125,
+      "learning_rate": 6.856256432000719e-09,
+      "loss": 0.0084,
+      "reward": 1.17421875,
+      "reward_std": 0.26185047421604396,
+      "rewards/accuracy_reward": 0.25546875,
+      "rewards/format_reward": 0.91875,
+      "step": 560
+    },
+    {
+      "completion_length": 263.44296875,
+      "epoch": 0.9982332155477032,
+      "grad_norm": 0.9917849444243174,
+      "kl": 0.215625,
+      "learning_rate": 1.904727299473219e-10,
+      "loss": 0.0086,
+      "reward": 1.159375,
+      "reward_std": 0.23422911586239933,
+      "rewards/accuracy_reward": 0.23671875,
+      "rewards/format_reward": 0.92265625,
+      "step": 565
+    },
+    {
+      "completion_length": 255.21180772781372,
+      "epoch": 1.0,
+      "kl": 0.172515869140625,
+      "reward": 1.15625,
+      "reward_std": 0.20992232067510486,
+      "rewards/accuracy_reward": 0.23046875,
+      "rewards/format_reward": 0.92578125,
+      "step": 566,
+      "total_flos": 0.0,
+      "train_loss": 38.52571455222895,
+      "train_runtime": 51743.6213,
+      "train_samples_per_second": 1.4,
+      "train_steps_per_second": 0.011
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 566,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a004457909a0a3ddb16baca252276c646e7508124d698483539f8424afc2fab
+size 7544