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 +1,67 @@
+---
+library_name: transformers
+model_name: Qwen2.5-0.5B-Open-R1-GRPO
+tags:
+- generated_from_trainer
+- trl
+- grpo
+licence: license
+---
+
+# Model Card for Qwen2.5-0.5B-Open-R1-GRPO
+
+This model is a fine-tuned version of [None](https://huggingface.co/None).
+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="jdqqjr/Qwen2.5-0.5B-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/xdrshjr/huggingface/runs/4qpc7qth) 
+
+
+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+cu121
+- 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": 0.4606501721456687,
+    "train_runtime": 51552.651,
+    "train_samples": 72441,
+    "train_samples_per_second": 0.281,
+    "train_steps_per_second": 0.023
+}

config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "data/Qwen2.5-0.5B-Open-R1-Distill",
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 896,
+  "initializer_range": 0.02,
+  "intermediate_size": 4864,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 21,
+  "model_type": "qwen2",
+  "num_attention_heads": 14,
+  "num_hidden_layers": 24,
+  "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:c830f69a3297fce594de0cbaafcc582b9ea7013e832c4c28a2fdfa28869f6480
+size 1260367448

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": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63a2951d5edfa5cc0a2346ef872f8c77a2920274cfc3b503b04e3799104dee80
+size 11422060

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": "<|im_end|>",
+  "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": 0.4606501721456687,
+    "train_runtime": 51552.651,
+    "train_samples": 72441,
+    "train_samples_per_second": 0.281,
+    "train_steps_per_second": 0.023
+}

trainer_state.json

@@ -0,0 +1,1692 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.20010767383111774,
+  "eval_steps": 200,
+  "global_step": 1208,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "completion_length": 896.090625,
+      "epoch": 0.0016565204787344183,
+      "grad_norm": 1.3095546960830688,
+      "kl": 0.0007995764414469401,
+      "learning_rate": 1.6528925619834712e-06,
+      "loss": 0.0,
+      "reward": 0.022916666666666665,
+      "reward_std": 0.05667823006709417,
+      "rewards/accuracy_reward": 0.020833333333333332,
+      "rewards/format_reward": 0.0020833333333333333,
+      "step": 10
+    },
+    {
+      "completion_length": 872.3708333333333,
+      "epoch": 0.0033130409574688366,
+      "grad_norm": 0.29464519023895264,
+      "kl": 0.0228486696879069,
+      "learning_rate": 3.3057851239669424e-06,
+      "loss": 0.0009,
+      "reward": 0.036458333333333336,
+      "reward_std": 0.07045734748244285,
+      "rewards/accuracy_reward": 0.027083333333333334,
+      "rewards/format_reward": 0.009375,
+      "step": 20
+    },
+    {
+      "completion_length": 868.2104166666667,
+      "epoch": 0.004969561436203255,
+      "grad_norm": 1.250253438949585,
+      "kl": 0.012819226582845051,
+      "learning_rate": 4.958677685950414e-06,
+      "loss": 0.0005,
+      "reward": 0.058333333333333334,
+      "reward_std": 0.10454692070682843,
+      "rewards/accuracy_reward": 0.04375,
+      "rewards/format_reward": 0.014583333333333334,
+      "step": 30
+    },
+    {
+      "completion_length": 352.371875,
+      "epoch": 0.006626081914937673,
+      "grad_norm": 3.138777256011963,
+      "kl": 0.19162521362304688,
+      "learning_rate": 6.611570247933885e-06,
+      "loss": 0.0077,
+      "reward": 0.54375,
+      "reward_std": 0.351862679173549,
+      "rewards/accuracy_reward": 0.058333333333333334,
+      "rewards/format_reward": 0.48541666666666666,
+      "step": 40
+    },
+    {
+      "completion_length": 269.03229166666665,
+      "epoch": 0.008282602393672091,
+      "grad_norm": 2.153762102127075,
+      "kl": 0.17170003255208333,
+      "learning_rate": 8.264462809917356e-06,
+      "loss": 0.0069,
+      "reward": 0.7614583333333333,
+      "reward_std": 0.29137441540757814,
+      "rewards/accuracy_reward": 0.027083333333333334,
+      "rewards/format_reward": 0.734375,
+      "step": 50
+    },
+    {
+      "completion_length": 201.54479166666667,
+      "epoch": 0.00993912287240651,
+      "grad_norm": 1.2917091846466064,
+      "kl": 0.23529866536458333,
+      "learning_rate": 9.917355371900828e-06,
+      "loss": 0.0094,
+      "reward": 0.8385416666666666,
+      "reward_std": 0.2966347888112068,
+      "rewards/accuracy_reward": 0.034375,
+      "rewards/format_reward": 0.8041666666666667,
+      "step": 60
+    },
+    {
+      "completion_length": 138.68958333333333,
+      "epoch": 0.011595643351140928,
+      "grad_norm": 1.463751196861267,
+      "kl": 0.35206705729166665,
+      "learning_rate": 1.1570247933884297e-05,
+      "loss": 0.0141,
+      "reward": 0.9666666666666667,
+      "reward_std": 0.19539905513326328,
+      "rewards/accuracy_reward": 0.03125,
+      "rewards/format_reward": 0.9354166666666667,
+      "step": 70
+    },
+    {
+      "completion_length": 103.84791666666666,
+      "epoch": 0.013252163829875346,
+      "grad_norm": 1.8092870712280273,
+      "kl": 0.5803548177083333,
+      "learning_rate": 1.322314049586777e-05,
+      "loss": 0.0232,
+      "reward": 0.9958333333333333,
+      "reward_std": 0.22554169197877247,
+      "rewards/accuracy_reward": 0.06041666666666667,
+      "rewards/format_reward": 0.9354166666666667,
+      "step": 80
+    },
+    {
+      "completion_length": 115.95208333333333,
+      "epoch": 0.014908684308609765,
+      "grad_norm": 1.7465665340423584,
+      "kl": 0.4664306640625,
+      "learning_rate": 1.487603305785124e-05,
+      "loss": 0.0187,
+      "reward": 1.009375,
+      "reward_std": 0.21032156000534694,
+      "rewards/accuracy_reward": 0.06979166666666667,
+      "rewards/format_reward": 0.9395833333333333,
+      "step": 90
+    },
+    {
+      "completion_length": 118.71041666666666,
+      "epoch": 0.016565204787344183,
+      "grad_norm": 3.3586018085479736,
+      "kl": 28.244026692708335,
+      "learning_rate": 1.652892561983471e-05,
+      "loss": 1.1292,
+      "reward": 1.0447916666666666,
+      "reward_std": 0.23326072320342064,
+      "rewards/accuracy_reward": 0.09895833333333333,
+      "rewards/format_reward": 0.9458333333333333,
+      "step": 100
+    },
+    {
+      "completion_length": 118.61875,
+      "epoch": 0.0182217252660786,
+      "grad_norm": 1268.205322265625,
+      "kl": 4.216959635416667,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.1686,
+      "reward": 0.9927083333333333,
+      "reward_std": 0.3241697837909063,
+      "rewards/accuracy_reward": 0.09479166666666666,
+      "rewards/format_reward": 0.8979166666666667,
+      "step": 110
+    },
+    {
+      "completion_length": 187.56145833333332,
+      "epoch": 0.01987824574481302,
+      "grad_norm": 42.72395324707031,
+      "kl": 11.572265625,
+      "learning_rate": 1.9834710743801656e-05,
+      "loss": 0.4629,
+      "reward": 0.859375,
+      "reward_std": 0.3210473525027434,
+      "rewards/accuracy_reward": 0.0375,
+      "rewards/format_reward": 0.821875,
+      "step": 120
+    },
+    {
+      "completion_length": 79.61979166666667,
+      "epoch": 0.021534766223547438,
+      "grad_norm": 2.5782203674316406,
+      "kl": 2.37890625,
+      "learning_rate": 1.9996617239944584e-05,
+      "loss": 0.0951,
+      "reward": 0.9458333333333333,
+      "reward_std": 0.14564209878444673,
+      "rewards/accuracy_reward": 0.008333333333333333,
+      "rewards/format_reward": 0.9375,
+      "step": 130
+    },
+    {
+      "completion_length": 63.088541666666664,
+      "epoch": 0.023191286702281856,
+      "grad_norm": 612.989501953125,
+      "kl": 72.36246744791667,
+      "learning_rate": 1.9984926686598405e-05,
+      "loss": 2.8849,
+      "reward": 0.95625,
+      "reward_std": 0.1277017446855704,
+      "rewards/accuracy_reward": 0.005208333333333333,
+      "rewards/format_reward": 0.9510416666666667,
+      "step": 140
+    },
+    {
+      "completion_length": 81.41354166666666,
+      "epoch": 0.024847807181016274,
+      "grad_norm": 2.788867473602295,
+      "kl": 1.7197916666666666,
+      "learning_rate": 1.996489633956445e-05,
+      "loss": 0.0688,
+      "reward": 0.946875,
+      "reward_std": 0.13180716559290886,
+      "rewards/accuracy_reward": 0.010416666666666666,
+      "rewards/format_reward": 0.9364583333333333,
+      "step": 150
+    },
+    {
+      "completion_length": 91.93541666666667,
+      "epoch": 0.026504327659750693,
+      "grad_norm": 8.987010955810547,
+      "kl": 1.2494954427083333,
+      "learning_rate": 1.9936542928956713e-05,
+      "loss": 0.05,
+      "reward": 0.9760416666666667,
+      "reward_std": 0.19224367886781693,
+      "rewards/accuracy_reward": 0.046875,
+      "rewards/format_reward": 0.9291666666666667,
+      "step": 160
+    },
+    {
+      "completion_length": 39.534375,
+      "epoch": 0.02816084813848511,
+      "grad_norm": 1.1727018356323242,
+      "kl": 173.3283203125,
+      "learning_rate": 1.9899890136631088e-05,
+      "loss": 6.9265,
+      "reward": 1.028125,
+      "reward_std": 0.11864924083153407,
+      "rewards/accuracy_reward": 0.051041666666666666,
+      "rewards/format_reward": 0.9770833333333333,
+      "step": 170
+    },
+    {
+      "completion_length": 33.384375,
+      "epoch": 0.02981736861721953,
+      "grad_norm": 1.362429141998291,
+      "kl": 1.3204427083333334,
+      "learning_rate": 1.9854968576405343e-05,
+      "loss": 0.0528,
+      "reward": 1.075,
+      "reward_std": 0.1386565258105596,
+      "rewards/accuracy_reward": 0.10416666666666667,
+      "rewards/format_reward": 0.9708333333333333,
+      "step": 180
+    },
+    {
+      "completion_length": 69.90416666666667,
+      "epoch": 0.03147388909595395,
+      "grad_norm": 110.89439392089844,
+      "kl": 1.244873046875,
+      "learning_rate": 1.980181576848931e-05,
+      "loss": 0.0498,
+      "reward": 1.0354166666666667,
+      "reward_std": 0.17048126459121704,
+      "rewards/accuracy_reward": 0.07604166666666666,
+      "rewards/format_reward": 0.959375,
+      "step": 190
+    },
+    {
+      "completion_length": 301.9614583333333,
+      "epoch": 0.033130409574688366,
+      "grad_norm": 4.626757621765137,
+      "kl": 1.0924153645833334,
+      "learning_rate": 1.974047610814655e-05,
+      "loss": 0.0437,
+      "reward": 0.7625,
+      "reward_std": 0.32025263756513594,
+      "rewards/accuracy_reward": 0.025,
+      "rewards/format_reward": 0.7375,
+      "step": 200
+    },
+    {
+      "epoch": 0.033130409574688366,
+      "eval_completion_length": 104.2790404040404,
+      "eval_kl": 0.7861624053030303,
+      "eval_loss": 0.03144485875964165,
+      "eval_reward": 1.0303030303030303,
+      "eval_reward_std": 0.17504648879320936,
+      "eval_rewards/accuracy_reward": 0.07575757575757576,
+      "eval_rewards/format_reward": 0.9545454545454546,
+      "eval_runtime": 117.4071,
+      "eval_samples_per_second": 0.843,
+      "eval_steps_per_second": 0.843,
+      "step": 200
+    },
+    {
+      "completion_length": 332.1666666666667,
+      "epoch": 0.03478693005342279,
+      "grad_norm": 31.273311614990234,
+      "kl": 1.1237955729166667,
+      "learning_rate": 1.967100082861375e-05,
+      "loss": 0.045,
+      "reward": 0.7010416666666667,
+      "reward_std": 0.3886319617430369,
+      "rewards/accuracy_reward": 0.026041666666666668,
+      "rewards/format_reward": 0.675,
+      "step": 210
+    },
+    {
+      "completion_length": 72.00208333333333,
+      "epoch": 0.0364434505321572,
+      "grad_norm": 19.649843215942383,
+      "kl": 1.6216471354166666,
+      "learning_rate": 1.9593447958308732e-05,
+      "loss": 0.0649,
+      "reward": 1.009375,
+      "reward_std": 0.11707024797797203,
+      "rewards/accuracy_reward": 0.03333333333333333,
+      "rewards/format_reward": 0.9760416666666667,
+      "step": 220
+    },
+    {
+      "completion_length": 78.76770833333333,
+      "epoch": 0.038099971010891624,
+      "grad_norm": 11.606145858764648,
+      "kl": 102.15354817708334,
+      "learning_rate": 1.9507882272362896e-05,
+      "loss": 4.0776,
+      "reward": 1.0166666666666666,
+      "reward_std": 0.1833244909842809,
+      "rewards/accuracy_reward": 0.06145833333333333,
+      "rewards/format_reward": 0.9552083333333333,
+      "step": 230
+    },
+    {
+      "completion_length": 138.13229166666667,
+      "epoch": 0.03975649148962604,
+      "grad_norm": 16.69879150390625,
+      "kl": 461.5420247395833,
+      "learning_rate": 1.9414375238518572e-05,
+      "loss": 18.4967,
+      "reward": 0.9645833333333333,
+      "reward_std": 0.19629337415099143,
+      "rewards/accuracy_reward": 0.0375,
+      "rewards/format_reward": 0.9270833333333334,
+      "step": 240
+    },
+    {
+      "completion_length": 204.47083333333333,
+      "epoch": 0.04141301196836046,
+      "grad_norm": 3.3681952953338623,
+      "kl": 1.022412109375,
+      "learning_rate": 1.9313004957436394e-05,
+      "loss": 0.0409,
+      "reward": 0.9229166666666667,
+      "reward_std": 0.25351568634311356,
+      "rewards/accuracy_reward": 0.036458333333333336,
+      "rewards/format_reward": 0.8864583333333333,
+      "step": 250
+    },
+    {
+      "completion_length": 113.05833333333334,
+      "epoch": 0.043069532447094876,
+      "grad_norm": 1.5134903192520142,
+      "kl": 17.833056640625,
+      "learning_rate": 1.920385609746264e-05,
+      "loss": 0.7128,
+      "reward": 1.0208333333333333,
+      "reward_std": 0.17660286352038385,
+      "rewards/accuracy_reward": 0.06041666666666667,
+      "rewards/format_reward": 0.9604166666666667,
+      "step": 260
+    },
+    {
+      "completion_length": 89.409375,
+      "epoch": 0.0447260529258293,
+      "grad_norm": 6.134641647338867,
+      "kl": 1.5666829427083333,
+      "learning_rate": 1.9087019823910995e-05,
+      "loss": 0.0626,
+      "reward": 1.0510416666666667,
+      "reward_std": 0.1831239640712738,
+      "rewards/accuracy_reward": 0.07916666666666666,
+      "rewards/format_reward": 0.971875,
+      "step": 270
+    },
+    {
+      "completion_length": 130.04479166666667,
+      "epoch": 0.04638257340456371,
+      "grad_norm": 12.951639175415039,
+      "kl": 350.734228515625,
+      "learning_rate": 1.8962593722917763e-05,
+      "loss": 14.0643,
+      "reward": 1.0083333333333333,
+      "reward_std": 0.2503954492509365,
+      "rewards/accuracy_reward": 0.07708333333333334,
+      "rewards/format_reward": 0.93125,
+      "step": 280
+    },
+    {
+      "completion_length": 115.55416666666666,
+      "epoch": 0.048039093883298134,
+      "grad_norm": 3.048227548599243,
+      "kl": 2.5464518229166666,
+      "learning_rate": 1.8830681719934197e-05,
+      "loss": 0.1018,
+      "reward": 0.96875,
+      "reward_std": 0.1739353731274605,
+      "rewards/accuracy_reward": 0.026041666666666668,
+      "rewards/format_reward": 0.9427083333333334,
+      "step": 290
+    },
+    {
+      "completion_length": 86.45729166666666,
+      "epoch": 0.04969561436203255,
+      "grad_norm": 3.096956729888916,
+      "kl": 2.845361328125,
+      "learning_rate": 1.8691393992923984e-05,
+      "loss": 0.1138,
+      "reward": 1.034375,
+      "reward_std": 0.1851226476331552,
+      "rewards/accuracy_reward": 0.06875,
+      "rewards/format_reward": 0.965625,
+      "step": 300
+    },
+    {
+      "completion_length": 98.94479166666666,
+      "epoch": 0.05135213484076697,
+      "grad_norm": 2.9729843139648438,
+      "kl": 1.6844075520833333,
+      "learning_rate": 1.8544846880338362e-05,
+      "loss": 0.0674,
+      "reward": 1.0385416666666667,
+      "reward_std": 0.20008668601512908,
+      "rewards/accuracy_reward": 0.075,
+      "rewards/format_reward": 0.9635416666666666,
+      "step": 310
+    },
+    {
+      "completion_length": 159.10729166666667,
+      "epoch": 0.053008655319501385,
+      "grad_norm": 13.849080085754395,
+      "kl": 10.700813802083333,
+      "learning_rate": 1.8391162783945805e-05,
+      "loss": 0.4282,
+      "reward": 0.971875,
+      "reward_std": 0.25769852722684544,
+      "rewards/accuracy_reward": 0.06666666666666667,
+      "rewards/format_reward": 0.9052083333333333,
+      "step": 320
+    },
+    {
+      "completion_length": 204.58125,
+      "epoch": 0.05466517579823581,
+      "grad_norm": 7.857761383056641,
+      "kl": 35.507259114583334,
+      "learning_rate": 1.8230470066597365e-05,
+      "loss": 1.4198,
+      "reward": 0.896875,
+      "reward_std": 0.31051936596632,
+      "rewards/accuracy_reward": 0.046875,
+      "rewards/format_reward": 0.85,
+      "step": 330
+    },
+    {
+      "completion_length": 117.10729166666667,
+      "epoch": 0.05632169627697022,
+      "grad_norm": 5.027503490447998,
+      "kl": 2.419173177083333,
+      "learning_rate": 1.806290294501309e-05,
+      "loss": 0.0968,
+      "reward": 1.0375,
+      "reward_std": 0.19644681960344315,
+      "rewards/accuracy_reward": 0.08229166666666667,
+      "rewards/format_reward": 0.9552083333333333,
+      "step": 340
+    },
+    {
+      "completion_length": 114.72916666666667,
+      "epoch": 0.057978216755704644,
+      "grad_norm": 10.025717735290527,
+      "kl": 2.844384765625,
+      "learning_rate": 1.7888601377679075e-05,
+      "loss": 0.1137,
+      "reward": 1.03125,
+      "reward_std": 0.20197176635265351,
+      "rewards/accuracy_reward": 0.07604166666666666,
+      "rewards/format_reward": 0.9552083333333333,
+      "step": 350
+    },
+    {
+      "completion_length": 113.32708333333333,
+      "epoch": 0.05963473723443906,
+      "grad_norm": 1.1121678352355957,
+      "kl": 3.6202799479166665,
+      "learning_rate": 1.770771094794878e-05,
+      "loss": 0.1447,
+      "reward": 0.9895833333333334,
+      "reward_std": 0.2017668180167675,
+      "rewards/accuracy_reward": 0.04791666666666667,
+      "rewards/format_reward": 0.9416666666666667,
+      "step": 360
+    },
+    {
+      "completion_length": 150.92291666666668,
+      "epoch": 0.06129125771317348,
+      "grad_norm": 10.448979377746582,
+      "kl": 7.117659505208334,
+      "learning_rate": 1.7520382742446208e-05,
+      "loss": 0.2847,
+      "reward": 0.9666666666666667,
+      "reward_std": 0.304207677145799,
+      "rewards/accuracy_reward": 0.07604166666666666,
+      "rewards/format_reward": 0.890625,
+      "step": 370
+    },
+    {
+      "completion_length": 86.54583333333333,
+      "epoch": 0.0629477781919079,
+      "grad_norm": 1.3881269693374634,
+      "kl": 16.287451171875,
+      "learning_rate": 1.732677322487258e-05,
+      "loss": 0.6512,
+      "reward": 1.0739583333333333,
+      "reward_std": 0.18988159050544104,
+      "rewards/accuracy_reward": 0.10625,
+      "rewards/format_reward": 0.9677083333333333,
+      "step": 380
+    },
+    {
+      "completion_length": 120.71875,
+      "epoch": 0.06460429867064231,
+      "grad_norm": 7.372952938079834,
+      "kl": 20.257828776041666,
+      "learning_rate": 1.712704410532185e-05,
+      "loss": 0.8119,
+      "reward": 1.0229166666666667,
+      "reward_std": 0.2337183562417825,
+      "rewards/accuracy_reward": 0.10416666666666667,
+      "rewards/format_reward": 0.91875,
+      "step": 390
+    },
+    {
+      "completion_length": 100.46354166666667,
+      "epoch": 0.06626081914937673,
+      "grad_norm": 1.1761277914047241,
+      "kl": 2.2524739583333333,
+      "learning_rate": 1.6921362205214212e-05,
+      "loss": 0.0901,
+      "reward": 1.0083333333333333,
+      "reward_std": 0.20387225871284803,
+      "rewards/accuracy_reward": 0.08125,
+      "rewards/format_reward": 0.9270833333333334,
+      "step": 400
+    },
+    {
+      "epoch": 0.06626081914937673,
+      "eval_completion_length": 66.43939393939394,
+      "eval_kl": 2.4309895833333335,
+      "eval_loss": 0.09729723632335663,
+      "eval_reward": 1.095959595959596,
+      "eval_reward_std": 0.15316122258552398,
+      "eval_rewards/accuracy_reward": 0.11868686868686869,
+      "eval_rewards/format_reward": 0.9772727272727273,
+      "eval_runtime": 64.9848,
+      "eval_samples_per_second": 1.523,
+      "eval_steps_per_second": 1.523,
+      "step": 400
+    },
+    {
+      "completion_length": 101.78020833333333,
+      "epoch": 0.06791733962811115,
+      "grad_norm": 6.115810871124268,
+      "kl": 2.965804036458333,
+      "learning_rate": 1.6709899317960486e-05,
+      "loss": 0.1186,
+      "reward": 0.996875,
+      "reward_std": 0.2588055488963922,
+      "rewards/accuracy_reward": 0.078125,
+      "rewards/format_reward": 0.91875,
+      "step": 410
+    },
+    {
+      "completion_length": 91.97916666666667,
+      "epoch": 0.06957386010684558,
+      "grad_norm": 2.940000295639038,
+      "kl": 2.007307942708333,
+      "learning_rate": 1.6492832065473648e-05,
+      "loss": 0.0803,
+      "reward": 1.0208333333333333,
+      "reward_std": 0.18291174744566283,
+      "rewards/accuracy_reward": 0.06770833333333333,
+      "rewards/format_reward": 0.953125,
+      "step": 420
+    },
+    {
+      "completion_length": 98.98854166666666,
+      "epoch": 0.07123038058557998,
+      "grad_norm": 4.15654993057251,
+      "kl": 1.5184733072916667,
+      "learning_rate": 1.627034175064744e-05,
+      "loss": 0.0607,
+      "reward": 1.0447916666666666,
+      "reward_std": 0.15590187733372052,
+      "rewards/accuracy_reward": 0.078125,
+      "rewards/format_reward": 0.9666666666666667,
+      "step": 430
+    },
+    {
+      "completion_length": 100.14479166666666,
+      "epoch": 0.0728869010643144,
+      "grad_norm": 1.6317799091339111,
+      "kl": 1.5751139322916667,
+      "learning_rate": 1.6042614205925246e-05,
+      "loss": 0.063,
+      "reward": 1.046875,
+      "reward_std": 0.182314832508564,
+      "rewards/accuracy_reward": 0.08541666666666667,
+      "rewards/format_reward": 0.9614583333333333,
+      "step": 440
+    },
+    {
+      "completion_length": 97.09791666666666,
+      "epoch": 0.07454342154304883,
+      "grad_norm": 5.319403171539307,
+      "kl": 1.705322265625,
+      "learning_rate": 1.580983963808571e-05,
+      "loss": 0.0682,
+      "reward": 1.0479166666666666,
+      "reward_std": 0.16928049996495248,
+      "rewards/accuracy_reward": 0.078125,
+      "rewards/format_reward": 0.9697916666666667,
+      "step": 450
+    },
+    {
+      "completion_length": 124.80208333333333,
+      "epoch": 0.07619994202178325,
+      "grad_norm": 1.9065351486206055,
+      "kl": 1.8480143229166666,
+      "learning_rate": 1.5572212469374725e-05,
+      "loss": 0.0739,
+      "reward": 0.99375,
+      "reward_std": 0.18316799898942313,
+      "rewards/accuracy_reward": 0.042708333333333334,
+      "rewards/format_reward": 0.9510416666666667,
+      "step": 460
+    },
+    {
+      "completion_length": 126.94375,
+      "epoch": 0.07785646250051766,
+      "grad_norm": 0.715796709060669,
+      "kl": 0.93486328125,
+      "learning_rate": 1.5329931175116542e-05,
+      "loss": 0.0374,
+      "reward": 0.9885416666666667,
+      "reward_std": 0.10139760226011277,
+      "rewards/accuracy_reward": 0.015625,
+      "rewards/format_reward": 0.9729166666666667,
+      "step": 470
+    },
+    {
+      "completion_length": 124.96979166666667,
+      "epoch": 0.07951298297925208,
+      "grad_norm": 1.0792052745819092,
+      "kl": 0.705078125,
+      "learning_rate": 1.5083198117939532e-05,
+      "loss": 0.0282,
+      "reward": 1.0114583333333333,
+      "reward_std": 0.1728811038037141,
+      "rewards/accuracy_reward": 0.051041666666666666,
+      "rewards/format_reward": 0.9604166666666667,
+      "step": 480
+    },
+    {
+      "completion_length": 184.52083333333334,
+      "epoch": 0.0811695034579865,
+      "grad_norm": 1.0610744953155518,
+      "kl": 0.8315104166666667,
+      "learning_rate": 1.4832219378755206e-05,
+      "loss": 0.0333,
+      "reward": 0.9708333333333333,
+      "reward_std": 0.27192530358831085,
+      "rewards/accuracy_reward": 0.06145833333333333,
+      "rewards/format_reward": 0.909375,
+      "step": 490
+    },
+    {
+      "completion_length": 133.02083333333334,
+      "epoch": 0.08282602393672092,
+      "grad_norm": 1.3058396577835083,
+      "kl": 0.82529296875,
+      "learning_rate": 1.4577204584631484e-05,
+      "loss": 0.033,
+      "reward": 1.0614583333333334,
+      "reward_std": 0.20004068240523337,
+      "rewards/accuracy_reward": 0.09583333333333334,
+      "rewards/format_reward": 0.965625,
+      "step": 500
+    },
+    {
+      "completion_length": 141.05729166666666,
+      "epoch": 0.08448254441545533,
+      "grad_norm": 6.294867038726807,
+      "kl": 1.4442057291666666,
+      "learning_rate": 1.431836673370417e-05,
+      "loss": 0.0578,
+      "reward": 1.0666666666666667,
+      "reward_std": 0.26950543373823166,
+      "rewards/accuracy_reward": 0.12395833333333334,
+      "rewards/format_reward": 0.9427083333333334,
+      "step": 510
+    },
+    {
+      "completion_length": 134.06770833333334,
+      "epoch": 0.08613906489418975,
+      "grad_norm": 0.7419238090515137,
+      "kl": 0.8756510416666666,
+      "learning_rate": 1.4055922017272718e-05,
+      "loss": 0.035,
+      "reward": 1.0447916666666666,
+      "reward_std": 0.16486839627226194,
+      "rewards/accuracy_reward": 0.08229166666666667,
+      "rewards/format_reward": 0.9625,
+      "step": 520
+    },
+    {
+      "completion_length": 165.38333333333333,
+      "epoch": 0.08779558537292417,
+      "grad_norm": 1.7814686298370361,
+      "kl": 0.7268717447916667,
+      "learning_rate": 1.3790089639228969e-05,
+      "loss": 0.0291,
+      "reward": 1.0364583333333333,
+      "reward_std": 0.2156364011267821,
+      "rewards/accuracy_reward": 0.0875,
+      "rewards/format_reward": 0.9489583333333333,
+      "step": 530
+    },
+    {
+      "completion_length": 957.828125,
+      "epoch": 0.0894521058516586,
+      "grad_norm": 0.439900279045105,
+      "kl": 0.5694742838541667,
+      "learning_rate": 1.3521091632969663e-05,
+      "loss": 0.0228,
+      "reward": 0.05625,
+      "reward_std": 0.052992693583170575,
+      "rewards/accuracy_reward": 0.05,
+      "rewards/format_reward": 0.00625,
+      "step": 540
+    },
+    {
+      "completion_length": 887.0697916666667,
+      "epoch": 0.091108626330393,
+      "grad_norm": 0.39931434392929077,
+      "kl": 0.5684895833333333,
+      "learning_rate": 1.324915267594563e-05,
+      "loss": 0.0227,
+      "reward": 0.06041666666666667,
+      "reward_std": 0.09925113444526991,
+      "rewards/accuracy_reward": 0.06041666666666667,
+      "rewards/format_reward": 0.0,
+      "step": 550
+    },
+    {
+      "completion_length": 980.0729166666666,
+      "epoch": 0.09276514680912742,
+      "grad_norm": 0.14813727140426636,
+      "kl": 0.41376953125,
+      "learning_rate": 1.2974499902002551e-05,
+      "loss": 0.0166,
+      "reward": 0.040625,
+      "reward_std": 0.0782427782813708,
+      "rewards/accuracy_reward": 0.040625,
+      "rewards/format_reward": 0.0,
+      "step": 560
+    },
+    {
+      "completion_length": 988.3635416666667,
+      "epoch": 0.09442166728786185,
+      "grad_norm": 0.2630240023136139,
+      "kl": 0.37579752604166666,
+      "learning_rate": 1.2697362711670087e-05,
+      "loss": 0.015,
+      "reward": 0.023958333333333335,
+      "reward_std": 0.05509858330090841,
+      "rewards/accuracy_reward": 0.023958333333333335,
+      "rewards/format_reward": 0.0,
+      "step": 570
+    },
+    {
+      "completion_length": 1013.55,
+      "epoch": 0.09607818776659627,
+      "grad_norm": 0.10274422913789749,
+      "kl": 0.2902099609375,
+      "learning_rate": 1.2417972580557733e-05,
+      "loss": 0.0116,
+      "reward": 0.019791666666666666,
+      "reward_std": 0.04580442284544309,
+      "rewards/accuracy_reward": 0.019791666666666666,
+      "rewards/format_reward": 0.0,
+      "step": 580
+    },
+    {
+      "completion_length": 992.0416666666666,
+      "epoch": 0.09773470824533069,
+      "grad_norm": 0.3809487819671631,
+      "kl": 0.2610270182291667,
+      "learning_rate": 1.213656286601753e-05,
+      "loss": 0.0104,
+      "reward": 0.035416666666666666,
+      "reward_std": 0.06183087428410848,
+      "rewards/accuracy_reward": 0.035416666666666666,
+      "rewards/format_reward": 0.0,
+      "step": 590
+    },
+    {
+      "completion_length": 985.4572916666667,
+      "epoch": 0.0993912287240651,
+      "grad_norm": 0.26037415862083435,
+      "kl": 0.27569173177083334,
+      "learning_rate": 1.1853368612235043e-05,
+      "loss": 0.011,
+      "reward": 0.053125,
+      "reward_std": 0.09055439804991086,
+      "rewards/accuracy_reward": 0.053125,
+      "rewards/format_reward": 0.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.0993912287240651,
+      "eval_completion_length": 957.5833333333334,
+      "eval_kl": 0.2780441130050505,
+      "eval_loss": 0.011122927069664001,
+      "eval_reward": 0.08080808080808081,
+      "eval_reward_std": 0.12153128030324223,
+      "eval_rewards/accuracy_reward": 0.08080808080808081,
+      "eval_rewards/format_reward": 0.0,
+      "eval_runtime": 442.5836,
+      "eval_samples_per_second": 0.224,
+      "eval_steps_per_second": 0.224,
+      "step": 600
+    },
+    {
+      "completion_length": 705.296875,
+      "epoch": 0.10104774920279952,
+      "grad_norm": 0.8704747557640076,
+      "kl": 0.36200358072916666,
+      "learning_rate": 1.1568626353911452e-05,
+      "loss": 0.0145,
+      "reward": 0.09791666666666667,
+      "reward_std": 0.14217745463053386,
+      "rewards/accuracy_reward": 0.09791666666666667,
+      "rewards/format_reward": 0.0,
+      "step": 610
+    },
+    {
+      "completion_length": 504.3802083333333,
+      "epoch": 0.10270426968153394,
+      "grad_norm": 0.6176109313964844,
+      "kl": 0.34889322916666665,
+      "learning_rate": 1.1282573918700695e-05,
+      "loss": 0.014,
+      "reward": 0.06145833333333333,
+      "reward_std": 0.09735760614275932,
+      "rewards/accuracy_reward": 0.06145833333333333,
+      "rewards/format_reward": 0.0,
+      "step": 620
+    },
+    {
+      "completion_length": 689.9822916666667,
+      "epoch": 0.10436079016026836,
+      "grad_norm": 0.24804024398326874,
+      "kl": 0.293115234375,
+      "learning_rate": 1.0995450228566689e-05,
+      "loss": 0.0117,
+      "reward": 0.046875,
+      "reward_std": 0.09290325219432513,
+      "rewards/accuracy_reward": 0.046875,
+      "rewards/format_reward": 0.0,
+      "step": 630
+    },
+    {
+      "completion_length": 658.1604166666667,
+      "epoch": 0.10601731063900277,
+      "grad_norm": 0.526967465877533,
+      "kl": 0.2737223307291667,
+      "learning_rate": 1.0707495100226547e-05,
+      "loss": 0.011,
+      "reward": 0.075,
+      "reward_std": 0.13351340889930724,
+      "rewards/accuracy_reward": 0.075,
+      "rewards/format_reward": 0.0,
+      "step": 640
+    },
+    {
+      "completion_length": 616.5104166666666,
+      "epoch": 0.10767383111773719,
+      "grad_norm": 0.8440618515014648,
+      "kl": 0.2568277994791667,
+      "learning_rate": 1.0418949044846461e-05,
+      "loss": 0.0103,
+      "reward": 0.10520833333333333,
+      "reward_std": 0.15076439355810484,
+      "rewards/accuracy_reward": 0.10520833333333333,
+      "rewards/format_reward": 0.0,
+      "step": 650
+    },
+    {
+      "completion_length": 738.7666666666667,
+      "epoch": 0.10933035159647161,
+      "grad_norm": 0.5366663932800293,
+      "kl": 0.3122965494791667,
+      "learning_rate": 1.0130053067157537e-05,
+      "loss": 0.0125,
+      "reward": 0.08854166666666667,
+      "reward_std": 0.14642114490270614,
+      "rewards/accuracy_reward": 0.08854166666666667,
+      "rewards/format_reward": 0.0,
+      "step": 660
+    },
+    {
+      "completion_length": 933.3291666666667,
+      "epoch": 0.11098687207520604,
+      "grad_norm": 0.5030396580696106,
+      "kl": 0.2572021484375,
+      "learning_rate": 9.841048464159424e-06,
+      "loss": 0.0103,
+      "reward": 0.1,
+      "reward_std": 0.15392434100310007,
+      "rewards/accuracy_reward": 0.1,
+      "rewards/format_reward": 0.0,
+      "step": 670
+    },
+    {
+      "completion_length": 811.2208333333333,
+      "epoch": 0.11264339255394044,
+      "grad_norm": 0.5776947140693665,
+      "kl": 0.20400390625,
+      "learning_rate": 9.552176623579793e-06,
+      "loss": 0.0082,
+      "reward": 0.09791666666666667,
+      "reward_std": 0.15655608375867208,
+      "rewards/accuracy_reward": 0.09791666666666667,
+      "rewards/format_reward": 0.0,
+      "step": 680
+    },
+    {
+      "completion_length": 628.0864583333333,
+      "epoch": 0.11429991303267487,
+      "grad_norm": 0.5586806535720825,
+      "kl": 0.22731119791666668,
+      "learning_rate": 9.26367882225805e-06,
+      "loss": 0.0091,
+      "reward": 0.128125,
+      "reward_std": 0.15729325438539188,
+      "rewards/accuracy_reward": 0.128125,
+      "rewards/format_reward": 0.0,
+      "step": 690
+    },
+    {
+      "completion_length": 664.8635416666667,
+      "epoch": 0.11595643351140929,
+      "grad_norm": 0.5535178780555725,
+      "kl": 0.255712890625,
+      "learning_rate": 8.975796024621668e-06,
+      "loss": 0.0102,
+      "reward": 0.12291666666666666,
+      "reward_std": 0.16865403428673745,
+      "rewards/accuracy_reward": 0.12291666666666666,
+      "rewards/format_reward": 0.0,
+      "step": 700
+    },
+    {
+      "completion_length": 707.1645833333333,
+      "epoch": 0.11761295399014371,
+      "grad_norm": 0.5410148501396179,
+      "kl": 0.30380859375,
+      "learning_rate": 8.688768681423455e-06,
+      "loss": 0.0122,
+      "reward": 0.07291666666666667,
+      "reward_std": 0.12176756635308265,
+      "rewards/accuracy_reward": 0.07291666666666667,
+      "rewards/format_reward": 0.0,
+      "step": 710
+    },
+    {
+      "completion_length": 662.3677083333333,
+      "epoch": 0.11926947446887812,
+      "grad_norm": 0.5554656386375427,
+      "kl": 0.2496826171875,
+      "learning_rate": 8.40283652890787e-06,
+      "loss": 0.01,
+      "reward": 0.12083333333333333,
+      "reward_std": 0.14512543057401975,
+      "rewards/accuracy_reward": 0.12083333333333333,
+      "rewards/format_reward": 0.0,
+      "step": 720
+    },
+    {
+      "completion_length": 700.853125,
+      "epoch": 0.12092599494761254,
+      "grad_norm": 0.7303351163864136,
+      "kl": 0.23452962239583333,
+      "learning_rate": 8.118238388574112e-06,
+      "loss": 0.0094,
+      "reward": 0.11145833333333334,
+      "reward_std": 0.17577177782853445,
+      "rewards/accuracy_reward": 0.11145833333333334,
+      "rewards/format_reward": 0.0,
+      "step": 730
+    },
+    {
+      "completion_length": 644.3854166666666,
+      "epoch": 0.12258251542634696,
+      "grad_norm": 0.5679881572723389,
+      "kl": 0.24859212239583334,
+      "learning_rate": 7.835211967703244e-06,
+      "loss": 0.0099,
+      "reward": 0.09895833333333333,
+      "reward_std": 0.15083361292878786,
+      "rewards/accuracy_reward": 0.09895833333333333,
+      "rewards/format_reward": 0.0,
+      "step": 740
+    },
+    {
+      "completion_length": 582.60625,
+      "epoch": 0.12423903590508138,
+      "grad_norm": 0.5085894465446472,
+      "kl": 0.2998046875,
+      "learning_rate": 7.553993660815967e-06,
+      "loss": 0.012,
+      "reward": 0.121875,
+      "reward_std": 0.1481029657026132,
+      "rewards/accuracy_reward": 0.121875,
+      "rewards/format_reward": 0.0,
+      "step": 750
+    },
+    {
+      "completion_length": 591.3708333333333,
+      "epoch": 0.1258955563838158,
+      "grad_norm": 0.5969461798667908,
+      "kl": 0.3027099609375,
+      "learning_rate": 7.274818352226829e-06,
+      "loss": 0.0121,
+      "reward": 0.11354166666666667,
+      "reward_std": 0.15378185287117957,
+      "rewards/accuracy_reward": 0.11354166666666667,
+      "rewards/format_reward": 0.0,
+      "step": 760
+    },
+    {
+      "completion_length": 655.2,
+      "epoch": 0.1275520768625502,
+      "grad_norm": 0.6580300331115723,
+      "kl": 0.2832112630208333,
+      "learning_rate": 6.997919219859861e-06,
+      "loss": 0.0113,
+      "reward": 0.140625,
+      "reward_std": 0.18134810104966165,
+      "rewards/accuracy_reward": 0.140625,
+      "rewards/format_reward": 0.0,
+      "step": 770
+    },
+    {
+      "completion_length": 643.0625,
+      "epoch": 0.12920859734128462,
+      "grad_norm": 0.2721022665500641,
+      "kl": 0.2739339192708333,
+      "learning_rate": 6.72352754048942e-06,
+      "loss": 0.011,
+      "reward": 0.11770833333333333,
+      "reward_std": 0.11839525749286016,
+      "rewards/accuracy_reward": 0.11770833333333333,
+      "rewards/format_reward": 0.0,
+      "step": 780
+    },
+    {
+      "completion_length": 639.4947916666666,
+      "epoch": 0.13086511782001906,
+      "grad_norm": 0.6956638693809509,
+      "kl": 0.26971028645833334,
+      "learning_rate": 6.451872496568961e-06,
+      "loss": 0.0108,
+      "reward": 0.12604166666666666,
+      "reward_std": 0.1566569499671459,
+      "rewards/accuracy_reward": 0.12604166666666666,
+      "rewards/format_reward": 0.0,
+      "step": 790
+    },
+    {
+      "completion_length": 603.303125,
+      "epoch": 0.13252163829875346,
+      "grad_norm": 0.40896186232566833,
+      "kl": 0.2611002604166667,
+      "learning_rate": 6.183180984809059e-06,
+      "loss": 0.0104,
+      "reward": 0.09583333333333334,
+      "reward_std": 0.12670654132962228,
+      "rewards/accuracy_reward": 0.09583333333333334,
+      "rewards/format_reward": 0.0,
+      "step": 800
+    },
+    {
+      "epoch": 0.13252163829875346,
+      "eval_completion_length": 547.8737373737374,
+      "eval_kl": 0.28197009154040403,
+      "eval_loss": 0.011283059604465961,
+      "eval_reward": 0.1275252525252525,
+      "eval_reward_std": 0.17803129312968013,
+      "eval_rewards/accuracy_reward": 0.1275252525252525,
+      "eval_rewards/format_reward": 0.0,
+      "eval_runtime": 356.4254,
+      "eval_samples_per_second": 0.278,
+      "eval_steps_per_second": 0.278,
+      "step": 800
+    },
+    {
+      "completion_length": 567.071875,
+      "epoch": 0.1341781587774879,
+      "grad_norm": 0.9343724250793457,
+      "kl": 0.32443033854166664,
+      "learning_rate": 5.9176774266645805e-06,
+      "loss": 0.013,
+      "reward": 0.103125,
+      "reward_std": 0.1444171652197838,
+      "rewards/accuracy_reward": 0.103125,
+      "rewards/format_reward": 0.0,
+      "step": 810
+    },
+    {
+      "completion_length": 568.4385416666667,
+      "epoch": 0.1358346792562223,
+      "grad_norm": 0.5075631737709045,
+      "kl": 0.3996175130208333,
+      "learning_rate": 5.655583580889276e-06,
+      "loss": 0.016,
+      "reward": 0.11041666666666666,
+      "reward_std": 0.15164470622936885,
+      "rewards/accuracy_reward": 0.11041666666666666,
+      "rewards/format_reward": 0.0,
+      "step": 820
+    },
+    {
+      "completion_length": 607.4395833333333,
+      "epoch": 0.13749119973495671,
+      "grad_norm": 0.8729663491249084,
+      "kl": 0.319921875,
+      "learning_rate": 5.39711835831436e-06,
+      "loss": 0.0128,
+      "reward": 0.09895833333333333,
+      "reward_std": 0.1480001377562682,
+      "rewards/accuracy_reward": 0.09895833333333333,
+      "rewards/format_reward": 0.0,
+      "step": 830
+    },
+    {
+      "completion_length": 681.0083333333333,
+      "epoch": 0.13914772021369115,
+      "grad_norm": 1.2170244455337524,
+      "kl": 0.37718098958333335,
+      "learning_rate": 5.1424976390057945e-06,
+      "loss": 0.0151,
+      "reward": 0.10625,
+      "reward_std": 0.1617087279756864,
+      "rewards/accuracy_reward": 0.10625,
+      "rewards/format_reward": 0.0,
+      "step": 840
+    },
+    {
+      "completion_length": 620.546875,
+      "epoch": 0.14080424069242556,
+      "grad_norm": 1.0550384521484375,
+      "kl": 0.25787760416666666,
+      "learning_rate": 4.891934091952986e-06,
+      "loss": 0.0103,
+      "reward": 0.10520833333333333,
+      "reward_std": 0.17377842714389166,
+      "rewards/accuracy_reward": 0.10520833333333333,
+      "rewards/format_reward": 0.0,
+      "step": 850
+    },
+    {
+      "completion_length": 608.7427083333333,
+      "epoch": 0.14246076117115997,
+      "grad_norm": 0.6739098429679871,
+      "kl": 0.2914225260416667,
+      "learning_rate": 4.645636997439468e-06,
+      "loss": 0.0117,
+      "reward": 0.08020833333333334,
+      "reward_std": 0.13246000508467357,
+      "rewards/accuracy_reward": 0.08020833333333334,
+      "rewards/format_reward": 0.0,
+      "step": 860
+    },
+    {
+      "completion_length": 602.1041666666666,
+      "epoch": 0.1441172816498944,
+      "grad_norm": 0.6213006973266602,
+      "kl": 0.3249593098958333,
+      "learning_rate": 4.403812072244007e-06,
+      "loss": 0.013,
+      "reward": 0.13958333333333334,
+      "reward_std": 0.190857628484567,
+      "rewards/accuracy_reward": 0.13958333333333334,
+      "rewards/format_reward": 0.0,
+      "step": 870
+    },
+    {
+      "completion_length": 599.3635416666667,
+      "epoch": 0.1457738021286288,
+      "grad_norm": 0.8679569363594055,
+      "kl": 0.28642171223958335,
+      "learning_rate": 4.166661297818037e-06,
+      "loss": 0.0115,
+      "reward": 0.10833333333333334,
+      "reward_std": 0.15785114417473475,
+      "rewards/accuracy_reward": 0.10833333333333334,
+      "rewards/format_reward": 0.0,
+      "step": 880
+    },
+    {
+      "completion_length": 631.9927083333333,
+      "epoch": 0.14743032260736325,
+      "grad_norm": 1.9089330434799194,
+      "kl": 0.4612223307291667,
+      "learning_rate": 3.9343827515830346e-06,
+      "loss": 0.0185,
+      "reward": 0.13020833333333334,
+      "reward_std": 0.1697061302761237,
+      "rewards/accuracy_reward": 0.13020833333333334,
+      "rewards/format_reward": 0.0,
+      "step": 890
+    },
+    {
+      "completion_length": 611.6770833333334,
+      "epoch": 0.14908684308609765,
+      "grad_norm": 0.7718966603279114,
+      "kl": 0.34353841145833336,
+      "learning_rate": 3.707170441488622e-06,
+      "loss": 0.0137,
+      "reward": 0.10416666666666667,
+      "reward_std": 0.14273599833250045,
+      "rewards/accuracy_reward": 0.10416666666666667,
+      "rewards/format_reward": 0.0,
+      "step": 900
+    },
+    {
+      "completion_length": 577.1041666666666,
+      "epoch": 0.15074336356483206,
+      "grad_norm": 2.250363349914551,
+      "kl": 0.4068603515625,
+      "learning_rate": 3.4852141439697264e-06,
+      "loss": 0.0163,
+      "reward": 0.15729166666666666,
+      "reward_std": 0.1889521633585294,
+      "rewards/accuracy_reward": 0.15729166666666666,
+      "rewards/format_reward": 0.0,
+      "step": 910
+    },
+    {
+      "completion_length": 557.6520833333333,
+      "epoch": 0.1523998840435665,
+      "grad_norm": 0.9748304486274719,
+      "kl": 0.39437662760416664,
+      "learning_rate": 3.2686992454379896e-06,
+      "loss": 0.0158,
+      "reward": 0.15208333333333332,
+      "reward_std": 0.1580947625140349,
+      "rewards/accuracy_reward": 0.15208333333333332,
+      "rewards/format_reward": 0.0,
+      "step": 920
+    },
+    {
+      "completion_length": 549.9833333333333,
+      "epoch": 0.1540564045223009,
+      "grad_norm": 3.7645211219787598,
+      "kl": 0.38448486328125,
+      "learning_rate": 3.0578065874399587e-06,
+      "loss": 0.0154,
+      "reward": 0.13541666666666666,
+      "reward_std": 0.1658468487362067,
+      "rewards/accuracy_reward": 0.13541666666666666,
+      "rewards/format_reward": 0.0,
+      "step": 930
+    },
+    {
+      "completion_length": 572.9354166666667,
+      "epoch": 0.1557129250010353,
+      "grad_norm": 0.6928295493125916,
+      "kl": 0.3385538736979167,
+      "learning_rate": 2.8527123156112614e-06,
+      "loss": 0.0135,
+      "reward": 0.14270833333333333,
+      "reward_std": 0.18846885015567144,
+      "rewards/accuracy_reward": 0.14270833333333333,
+      "rewards/format_reward": 0.0,
+      "step": 940
+    },
+    {
+      "completion_length": 568.7708333333334,
+      "epoch": 0.15736944547976975,
+      "grad_norm": 1.5970089435577393,
+      "kl": 0.26595865885416664,
+      "learning_rate": 2.6535877325530412e-06,
+      "loss": 0.0106,
+      "reward": 0.13333333333333333,
+      "reward_std": 0.1627914269765218,
+      "rewards/accuracy_reward": 0.13333333333333333,
+      "rewards/format_reward": 0.0,
+      "step": 950
+    },
+    {
+      "completion_length": 590.0510416666667,
+      "epoch": 0.15902596595850416,
+      "grad_norm": 1.9767380952835083,
+      "kl": 0.35971272786458336,
+      "learning_rate": 2.460599154753435e-06,
+      "loss": 0.0144,
+      "reward": 0.134375,
+      "reward_std": 0.1874842757980029,
+      "rewards/accuracy_reward": 0.134375,
+      "rewards/format_reward": 0.0,
+      "step": 960
+    },
+    {
+      "completion_length": 633.08125,
+      "epoch": 0.1606824864372386,
+      "grad_norm": 2.3113903999328613,
+      "kl": 0.5449422200520834,
+      "learning_rate": 2.2739077736736848e-06,
+      "loss": 0.0218,
+      "reward": 0.11666666666666667,
+      "reward_std": 0.17836359639962515,
+      "rewards/accuracy_reward": 0.11666666666666667,
+      "rewards/format_reward": 0.0,
+      "step": 970
+    },
+    {
+      "completion_length": 606.3760416666667,
+      "epoch": 0.162339006915973,
+      "grad_norm": 6.433048725128174,
+      "kl": 0.45375162760416665,
+      "learning_rate": 2.0936695211148274e-06,
+      "loss": 0.0181,
+      "reward": 0.14583333333333334,
+      "reward_std": 0.19383215804894766,
+      "rewards/accuracy_reward": 0.14583333333333334,
+      "rewards/format_reward": 0.0,
+      "step": 980
+    },
+    {
+      "completion_length": 616.515625,
+      "epoch": 0.1639955273947074,
+      "grad_norm": 2.6663458347320557,
+      "kl": 0.7337646484375,
+      "learning_rate": 1.9200349389774842e-06,
+      "loss": 0.0294,
+      "reward": 0.15416666666666667,
+      "reward_std": 0.20536668300628663,
+      "rewards/accuracy_reward": 0.15416666666666667,
+      "rewards/format_reward": 0.0,
+      "step": 990
+    },
+    {
+      "completion_length": 566.825,
+      "epoch": 0.16565204787344184,
+      "grad_norm": 1.0958058834075928,
+      "kl": 0.49773763020833334,
+      "learning_rate": 1.7531490535234974e-06,
+      "loss": 0.0199,
+      "reward": 0.16354166666666667,
+      "reward_std": 0.18386900251110394,
+      "rewards/accuracy_reward": 0.16354166666666667,
+      "rewards/format_reward": 0.0,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16565204787344184,
+      "eval_completion_length": 579.4494949494949,
+      "eval_kl": 0.3323419744318182,
+      "eval_loss": 0.013291975483298302,
+      "eval_reward": 0.1691919191919192,
+      "eval_reward_std": 0.21246563786208028,
+      "eval_rewards/accuracy_reward": 0.1691919191919192,
+      "eval_rewards/format_reward": 0.0,
+      "eval_runtime": 351.9148,
+      "eval_samples_per_second": 0.281,
+      "eval_steps_per_second": 0.281,
+      "step": 1000
+    },
+    {
+      "completion_length": 636.7666666666667,
+      "epoch": 0.16730856835217625,
+      "grad_norm": 0.9040474891662598,
+      "kl": 0.33194580078125,
+      "learning_rate": 1.5931512542444216e-06,
+      "loss": 0.0133,
+      "reward": 0.14479166666666668,
+      "reward_std": 0.16781194806098937,
+      "rewards/accuracy_reward": 0.14479166666666668,
+      "rewards/format_reward": 0.0,
+      "step": 1010
+    },
+    {
+      "completion_length": 582.3885416666667,
+      "epoch": 0.16896508883091066,
+      "grad_norm": 1.7288283109664917,
+      "kl": 0.38011067708333335,
+      "learning_rate": 1.440175177438089e-06,
+      "loss": 0.0152,
+      "reward": 0.15208333333333332,
+      "reward_std": 0.19551267102360725,
+      "rewards/accuracy_reward": 0.15208333333333332,
+      "rewards/format_reward": 0.0,
+      "step": 1020
+    },
+    {
+      "completion_length": 594.2979166666667,
+      "epoch": 0.1706216093096451,
+      "grad_norm": 1.6643884181976318,
+      "kl": 0.2834676106770833,
+      "learning_rate": 1.2943485945904344e-06,
+      "loss": 0.0113,
+      "reward": 0.13229166666666667,
+      "reward_std": 0.15322435299555462,
+      "rewards/accuracy_reward": 0.13229166666666667,
+      "rewards/format_reward": 0.0,
+      "step": 1030
+    },
+    {
+      "completion_length": 596.1239583333333,
+      "epoch": 0.1722781297883795,
+      "grad_norm": 1.9393773078918457,
+      "kl": 0.3358357747395833,
+      "learning_rate": 1.1557933056558612e-06,
+      "loss": 0.0134,
+      "reward": 0.16458333333333333,
+      "reward_std": 0.20839111184080442,
+      "rewards/accuracy_reward": 0.16458333333333333,
+      "rewards/format_reward": 0.0,
+      "step": 1040
+    },
+    {
+      "completion_length": 609.6333333333333,
+      "epoch": 0.17393465026711394,
+      "grad_norm": 4.972652912139893,
+      "kl": 0.5623616536458333,
+      "learning_rate": 1.0246250373252454e-06,
+      "loss": 0.0225,
+      "reward": 0.14479166666666668,
+      "reward_std": 0.19776065895954767,
+      "rewards/accuracy_reward": 0.14479166666666668,
+      "rewards/format_reward": 0.0,
+      "step": 1050
+    },
+    {
+      "completion_length": 652.096875,
+      "epoch": 0.17559117074584835,
+      "grad_norm": 2.3235297203063965,
+      "kl": 0.5280476888020833,
+      "learning_rate": 9.009533463665643e-07,
+      "loss": 0.0211,
+      "reward": 0.12708333333333333,
+      "reward_std": 0.17044669638077417,
+      "rewards/accuracy_reward": 0.12708333333333333,
+      "rewards/format_reward": 0.0,
+      "step": 1060
+    },
+    {
+      "completion_length": 619.540625,
+      "epoch": 0.17724769122458275,
+      "grad_norm": 1.0357528924942017,
+      "kl": 0.4579671223958333,
+      "learning_rate": 7.848815281188782e-07,
+      "loss": 0.0183,
+      "reward": 0.11875,
+      "reward_std": 0.14704825406273206,
+      "rewards/accuracy_reward": 0.11875,
+      "rewards/format_reward": 0.0,
+      "step": 1070
+    },
+    {
+      "completion_length": 610.2885416666667,
+      "epoch": 0.1789042117033172,
+      "grad_norm": 1.4905515909194946,
+      "kl": 0.38160807291666665,
+      "learning_rate": 6.765065302161111e-07,
+      "loss": 0.0153,
+      "reward": 0.14791666666666667,
+      "reward_std": 0.15851844151814778,
+      "rewards/accuracy_reward": 0.14791666666666667,
+      "rewards/format_reward": 0.0,
+      "step": 1080
+    },
+    {
+      "completion_length": 589.303125,
+      "epoch": 0.1805607321820516,
+      "grad_norm": 0.9482324123382568,
+      "kl": 0.4548258463541667,
+      "learning_rate": 5.759188716126606e-07,
+      "loss": 0.0182,
+      "reward": 0.153125,
+      "reward_std": 0.1988860237101714,
+      "rewards/accuracy_reward": 0.153125,
+      "rewards/format_reward": 0.0,
+      "step": 1090
+    },
+    {
+      "completion_length": 646.6270833333333,
+      "epoch": 0.182217252660786,
+      "grad_norm": 1.8010622262954712,
+      "kl": 0.6629964192708333,
+      "learning_rate": 4.832025669785012e-07,
+      "loss": 0.0265,
+      "reward": 0.11979166666666667,
+      "reward_std": 0.16462505708138148,
+      "rewards/accuracy_reward": 0.11979166666666667,
+      "rewards/format_reward": 0.0,
+      "step": 1100
+    },
+    {
+      "completion_length": 628.3989583333333,
+      "epoch": 0.18387377313952044,
+      "grad_norm": 1.6385632753372192,
+      "kl": 0.5240641276041667,
+      "learning_rate": 3.9843505652690793e-07,
+      "loss": 0.021,
+      "reward": 0.1125,
+      "reward_std": 0.16707373360792796,
+      "rewards/accuracy_reward": 0.1125,
+      "rewards/format_reward": 0.0,
+      "step": 1110
+    },
+    {
+      "completion_length": 647.1125,
+      "epoch": 0.18553029361825485,
+      "grad_norm": 0.7716726064682007,
+      "kl": 0.47424723307291666,
+      "learning_rate": 3.2168714133341993e-07,
+      "loss": 0.019,
+      "reward": 0.09791666666666667,
+      "reward_std": 0.15301133766770364,
+      "rewards/accuracy_reward": 0.09791666666666667,
+      "rewards/format_reward": 0.0,
+      "step": 1120
+    },
+    {
+      "completion_length": 644.2041666666667,
+      "epoch": 0.18718681409698928,
+      "grad_norm": 0.650982677936554,
+      "kl": 0.4752197265625,
+      "learning_rate": 2.530229242000681e-07,
+      "loss": 0.019,
+      "reward": 0.10520833333333333,
+      "reward_std": 0.15487387304504713,
+      "rewards/accuracy_reward": 0.10520833333333333,
+      "rewards/format_reward": 0.0,
+      "step": 1130
+    },
+    {
+      "completion_length": 628.2083333333334,
+      "epoch": 0.1888433345757237,
+      "grad_norm": 1.5503145456314087,
+      "kl": 0.39764811197916666,
+      "learning_rate": 1.9249975611425498e-07,
+      "loss": 0.0159,
+      "reward": 0.16458333333333333,
+      "reward_std": 0.20950584759314855,
+      "rewards/accuracy_reward": 0.16458333333333333,
+      "rewards/format_reward": 0.0,
+      "step": 1140
+    },
+    {
+      "completion_length": 630.378125,
+      "epoch": 0.1904998550544581,
+      "grad_norm": 0.8726694583892822,
+      "kl": 0.40301920572916666,
+      "learning_rate": 1.4016818834701297e-07,
+      "loss": 0.0161,
+      "reward": 0.1375,
+      "reward_std": 0.16949546659986178,
+      "rewards/accuracy_reward": 0.1375,
+      "rewards/format_reward": 0.0,
+      "step": 1150
+    },
+    {
+      "completion_length": 617.0010416666667,
+      "epoch": 0.19215637553319254,
+      "grad_norm": 1.6046355962753296,
+      "kl": 0.5024739583333333,
+      "learning_rate": 9.607193023064121e-08,
+      "loss": 0.0201,
+      "reward": 0.16145833333333334,
+      "reward_std": 0.21067009270191192,
+      "rewards/accuracy_reward": 0.16145833333333334,
+      "rewards/format_reward": 0.0,
+      "step": 1160
+    },
+    {
+      "completion_length": 592.840625,
+      "epoch": 0.19381289601192694,
+      "grad_norm": 1.6286725997924805,
+      "kl": 0.4784464518229167,
+      "learning_rate": 6.024781265099866e-08,
+      "loss": 0.0191,
+      "reward": 0.17291666666666666,
+      "reward_std": 0.20954511786500613,
+      "rewards/accuracy_reward": 0.17291666666666666,
+      "rewards/format_reward": 0.0,
+      "step": 1170
+    },
+    {
+      "completion_length": 614.5677083333334,
+      "epoch": 0.19546941649066138,
+      "grad_norm": 1.2413487434387207,
+      "kl": 0.50418701171875,
+      "learning_rate": 3.2725757284932655e-08,
+      "loss": 0.0202,
+      "reward": 0.12604166666666666,
+      "reward_std": 0.16374331067005793,
+      "rewards/accuracy_reward": 0.12604166666666666,
+      "rewards/format_reward": 0.0,
+      "step": 1180
+    },
+    {
+      "completion_length": 619.15625,
+      "epoch": 0.1971259369693958,
+      "grad_norm": 1.2318090200424194,
+      "kl": 0.5339396158854167,
+      "learning_rate": 1.3528751608548052e-08,
+      "loss": 0.0214,
+      "reward": 0.13958333333333334,
+      "reward_std": 0.15883336688081423,
+      "rewards/accuracy_reward": 0.13854166666666667,
+      "rewards/format_reward": 0.0010416666666666667,
+      "step": 1190
+    },
+    {
+      "completion_length": 618.125,
+      "epoch": 0.1987824574481302,
+      "grad_norm": 3.1665122509002686,
+      "kl": 0.5285725911458333,
+      "learning_rate": 2.67282969718341e-09,
+      "loss": 0.0211,
+      "reward": 0.15,
+      "reward_std": 0.16763123348355294,
+      "rewards/accuracy_reward": 0.15,
+      "rewards/format_reward": 0.0,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1987824574481302,
+      "eval_completion_length": 617.9343434343434,
+      "eval_kl": 0.6195154671717171,
+      "eval_loss": 0.02479601837694645,
+      "eval_reward": 0.13636363636363635,
+      "eval_reward_std": 0.1822404096824954,
+      "eval_rewards/accuracy_reward": 0.1351010101010101,
+      "eval_rewards/format_reward": 0.0012626262626262627,
+      "eval_runtime": 388.1413,
+      "eval_samples_per_second": 0.255,
+      "eval_steps_per_second": 0.255,
+      "step": 1200
+    },
+    {
+      "completion_length": 621.3216145833334,
+      "epoch": 0.20010767383111774,
+      "kl": 0.4912109375,
+      "reward": 0.14583333333333334,
+      "reward_std": 0.22295449549953142,
+      "rewards/accuracy_reward": 0.14583333333333334,
+      "rewards/format_reward": 0.0,
+      "step": 1208,
+      "total_flos": 0.0,
+      "train_loss": 0.4606501721456687,
+      "train_runtime": 51552.651,
+      "train_samples_per_second": 0.281,
+      "train_steps_per_second": 0.023
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1208,
+  "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": 1,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:434ab1c6aff082f9cde1462efffe8ee2b700ce062e566e6877a62e3ea7f25225
+size 7224