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="hwang595/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/hongyi-wang/huggingface/runs/5prysxfx) 
+
+
+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": 0.9507393147195028,
+    "train_runtime": 22343.714,
+    "train_samples": 72441,
+    "train_samples_per_second": 3.242,
+    "train_steps_per_second": 0.029
+}

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 sha256:848dcb1c2e492a16f120a5ec905625dbe0ea65801e9bc9981a4383f5c58ca49d
+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": 0.9507393147195028,
+    "train_runtime": 22343.714,
+    "train_samples": 72441,
+    "train_samples_per_second": 3.242,
+    "train_steps_per_second": 0.029
+}

trainer_state.json

@@ -0,0 +1,2081 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9987438399845395,
+  "eval_steps": 100,
+  "global_step": 646,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "completion_length": 401.1393030166626,
+      "epoch": 0.007730215479756498,
+      "grad_norm": 0.7982586536355022,
+      "kl": 0.00011790990829467774,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 0.0,
+      "reward": 0.11296463984181174,
+      "reward_std": 0.4115350260399282,
+      "rewards/accuracy_reward": 0.15892857974395155,
+      "rewards/cosine_scaled_reward": -0.10816632110727369,
+      "rewards/format_reward": 0.01785714365541935,
+      "rewards/reasoning_steps_reward": 0.044345241598784925,
+      "step": 5
+    },
+    {
+      "completion_length": 359.26787223815916,
+      "epoch": 0.015460430959512996,
+      "grad_norm": 17.031780484117963,
+      "kl": 0.03648009300231934,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.0015,
+      "reward": 0.09043576336698607,
+      "reward_std": 0.4158310419879854,
+      "rewards/accuracy_reward": 0.13035714905709028,
+      "rewards/cosine_scaled_reward": -0.10301662639685674,
+      "rewards/format_reward": 0.03750000176951289,
+      "rewards/reasoning_steps_reward": 0.025595240062102675,
+      "step": 10
+    },
+    {
+      "completion_length": 168.35357909202577,
+      "epoch": 0.023190646439269495,
+      "grad_norm": 2.2889296772224683,
+      "kl": 0.1260986328125,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.005,
+      "reward": 0.06340085107949563,
+      "reward_std": 0.26944970826152714,
+      "rewards/accuracy_reward": 0.06964286081492901,
+      "rewards/cosine_scaled_reward": -0.05207534619839862,
+      "rewards/format_reward": 0.041071430407464506,
+      "rewards/reasoning_steps_reward": 0.004761905176565051,
+      "step": 15
+    },
+    {
+      "completion_length": 103.71071922779083,
+      "epoch": 0.03092086191902599,
+      "grad_norm": 2.061644058453735,
+      "kl": 0.184234619140625,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.0074,
+      "reward": 0.08586497388314455,
+      "reward_std": 0.21096171860699542,
+      "rewards/accuracy_reward": 0.058928574435412884,
+      "rewards/cosine_scaled_reward": -0.013242176159837981,
+      "rewards/format_reward": 0.03392857303842902,
+      "rewards/reasoning_steps_reward": 0.006250000465661288,
+      "step": 20
+    },
+    {
+      "completion_length": 110.00357648134232,
+      "epoch": 0.03865107739878249,
+      "grad_norm": 2.8992544505659312,
+      "kl": 0.7248291015625,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.029,
+      "reward": 0.12482290570333135,
+      "reward_std": 0.21813481883291388,
+      "rewards/accuracy_reward": 0.07589286109432578,
+      "rewards/cosine_scaled_reward": 0.017977662503835746,
+      "rewards/format_reward": 0.01428571492433548,
+      "rewards/reasoning_steps_reward": 0.016666667861863972,
+      "step": 25
+    },
+    {
+      "completion_length": 45.733930891752244,
+      "epoch": 0.04638129287853899,
+      "grad_norm": 11.431757181071006,
+      "kl": 2.81328125,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.1125,
+      "reward": 0.06212213041180803,
+      "reward_std": 0.1072377573115773,
+      "rewards/accuracy_reward": 0.03392857294529676,
+      "rewards/cosine_scaled_reward": 0.012419746899831807,
+      "rewards/format_reward": 0.00714285746216774,
+      "rewards/reasoning_steps_reward": 0.008630953030660748,
+      "step": 30
+    },
+    {
+      "completion_length": 9.464286106824876,
+      "epoch": 0.054111508358295486,
+      "grad_norm": 0.5755556102272293,
+      "kl": 6.241015625,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.2497,
+      "reward": 0.015272756675949494,
+      "reward_std": 0.024263933256418114,
+      "rewards/accuracy_reward": 0.007142857555299998,
+      "rewards/cosine_scaled_reward": 0.004558471292239119,
+      "rewards/format_reward": 0.001785714365541935,
+      "rewards/reasoning_steps_reward": 0.0017857144121080637,
+      "step": 35
+    },
+    {
+      "completion_length": 6.6812501952052115,
+      "epoch": 0.06184172383805198,
+      "grad_norm": 5.133782054419905,
+      "kl": 133.16796875,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 5.3339,
+      "reward": 0.006814662339468214,
+      "reward_std": 0.006255550790274356,
+      "rewards/accuracy_reward": 0.00357142873108387,
+      "rewards/cosine_scaled_reward": 0.0017551381433008827,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.001488095335662365,
+      "step": 40
+    },
+    {
+      "completion_length": 3.806250123679638,
+      "epoch": 0.06957193931780849,
+      "grad_norm": 5.446995678880181,
+      "kl": 12.984375,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.5193,
+      "reward": -0.0006268640601774678,
+      "reward_std": 0.0008444518398732726,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0006268640835514816,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 45
+    },
+    {
+      "completion_length": 1.5669643387198449,
+      "epoch": 0.07730215479756498,
+      "grad_norm": 2.3394145041845285,
+      "kl": 21.7921875,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.872,
+      "reward": -1.1912497865296245e-05,
+      "reward_std": 9.54106117774245e-06,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -1.1912497842558879e-05,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 50
+    },
+    {
+      "completion_length": 1.3491071730852127,
+      "epoch": 0.08503237027732148,
+      "grad_norm": 0.7479018151097766,
+      "kl": 24.3375,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.9739,
+      "reward": -7.016389770342357e-06,
+      "reward_std": 8.243451605949304e-06,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -7.016389770342357e-06,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 55
+    },
+    {
+      "completion_length": 1.4455357551574708,
+      "epoch": 0.09276258575707798,
+      "grad_norm": 1.2382907375250831,
+      "kl": 22.5875,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.9036,
+      "reward": -9.098166776766447e-06,
+      "reward_std": 8.897681084363285e-06,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -9.098166776766447e-06,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 60
+    },
+    {
+      "completion_length": 3.0062501400709154,
+      "epoch": 0.10049280123683448,
+      "grad_norm": 0.9000789032207113,
+      "kl": 19.11015625,
+      "learning_rate": 2e-05,
+      "loss": 0.7642,
+      "reward": -0.000532294634598518,
+      "reward_std": 0.0007382091938138658,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0005322946360309721,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 65
+    },
+    {
+      "completion_length": 16.82946523576975,
+      "epoch": 0.10822301671659097,
+      "grad_norm": 2.019009108914015,
+      "kl": 23.421875,
+      "learning_rate": 1.999634547413886e-05,
+      "loss": 0.9368,
+      "reward": -0.005337684503536088,
+      "reward_std": 0.006817118118553012,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005337684468702442,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 70
+    },
+    {
+      "completion_length": 61.51428896188736,
+      "epoch": 0.11595323219634747,
+      "grad_norm": 0.765391154663079,
+      "kl": 22.9328125,
+      "learning_rate": 1.9985384567667278e-05,
+      "loss": 0.917,
+      "reward": -0.018308725309964302,
+      "reward_std": 0.025135993797334777,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.018308725309600503,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 75
+    },
+    {
+      "completion_length": 14.856251035630702,
+      "epoch": 0.12368344767610397,
+      "grad_norm": 0.5766637804649585,
+      "kl": 25.190625,
+      "learning_rate": 1.9967125291968495e-05,
+      "loss": 1.0076,
+      "reward": -0.004915532504958265,
+      "reward_std": 0.006951612820307674,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004915532504958265,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 80
+    },
+    {
+      "completion_length": 63.7750042244792,
+      "epoch": 0.13141366315586048,
+      "grad_norm": 0.3014979326507053,
+      "kl": 23.484375,
+      "learning_rate": 1.9941580992841562e-05,
+      "loss": 0.9393,
+      "reward": -0.016173791494043143,
+      "reward_std": 0.022050836245387017,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.01617379151732621,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 85
+    },
+    {
+      "completion_length": 18.968751035630703,
+      "epoch": 0.13914387863561697,
+      "grad_norm": 12.629651804585807,
+      "kl": 24.29375,
+      "learning_rate": 1.990877034074683e-05,
+      "loss": 0.9718,
+      "reward": -0.01146240434562742,
+      "reward_std": 0.01428736164973543,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.011462404392193547,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 90
+    },
+    {
+      "completion_length": 23.74196539670229,
+      "epoch": 0.14687409411537347,
+      "grad_norm": 7.695576654756108,
+      "kl": 22.0734375,
+      "learning_rate": 1.9868717317159617e-05,
+      "loss": 0.8824,
+      "reward": -0.02141171331331302,
+      "reward_std": 0.02541972674035673,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.021411713522860597,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 95
+    },
+    {
+      "completion_length": 5.691964483261108,
+      "epoch": 0.15460430959512997,
+      "grad_norm": 0.21802641340099765,
+      "kl": 25.30625,
+      "learning_rate": 1.9821451197042028e-05,
+      "loss": 1.0123,
+      "reward": -0.0023316225950111404,
+      "reward_std": 0.0032893259587581268,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0023316225950111404,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.15460430959512997,
+      "eval_completion_length": 5.566964626312256,
+      "eval_kl": 25.59375,
+      "eval_loss": 1.0240648984909058,
+      "eval_reward": -0.00207493151538074,
+      "eval_reward_std": 0.0029343964997678995,
+      "eval_rewards/accuracy_reward": 0.0,
+      "eval_rewards/cosine_scaled_reward": -0.00207493151538074,
+      "eval_rewards/format_reward": 0.0,
+      "eval_rewards/reasoning_steps_reward": 0.0,
+      "eval_runtime": 5.5856,
+      "eval_samples_per_second": 17.724,
+      "eval_steps_per_second": 0.716,
+      "step": 100
+    },
+    {
+      "completion_length": 123.02321983426809,
+      "epoch": 0.16233452507488647,
+      "grad_norm": 11.539029848998233,
+      "kl": 20.16171875,
+      "learning_rate": 1.9767006527445728e-05,
+      "loss": 0.8064,
+      "reward": -0.0462604992452043,
+      "reward_std": 0.05010210272388349,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.04626049910550591,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 105
+    },
+    {
+      "completion_length": 28.171430292725564,
+      "epoch": 0.17006474055464296,
+      "grad_norm": 0.3998672476939027,
+      "kl": 24.53125,
+      "learning_rate": 1.9705423102261324e-05,
+      "loss": 0.9813,
+      "reward": -0.008159401490456731,
+      "reward_std": 0.010330504266585194,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.008159401537022859,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 110
+    },
+    {
+      "completion_length": 60.81071784794331,
+      "epoch": 0.17779495603439946,
+      "grad_norm": 0.7363756914291658,
+      "kl": 23.45,
+      "learning_rate": 1.9636745933132807e-05,
+      "loss": 0.938,
+      "reward": -0.021001891322873688,
+      "reward_std": 0.028707678055423003,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.021001891322873688,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 115
+    },
+    {
+      "completion_length": 27.139287193119525,
+      "epoch": 0.18552517151415596,
+      "grad_norm": 0.9014468078078542,
+      "kl": 24.5640625,
+      "learning_rate": 1.956102521655831e-05,
+      "loss": 0.9826,
+      "reward": -0.014393773931152598,
+      "reward_std": 0.019338861525009854,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.014393773931152598,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 120
+    },
+    {
+      "completion_length": 33.95268049240112,
+      "epoch": 0.19325538699391245,
+      "grad_norm": 0.542203001284216,
+      "kl": 23.9765625,
+      "learning_rate": 1.9478316297201218e-05,
+      "loss": 0.959,
+      "reward": -0.016732960560364774,
+      "reward_std": 0.023507376017472836,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.016732960560364774,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 125
+    },
+    {
+      "completion_length": 50.27500329464674,
+      "epoch": 0.20098560247366895,
+      "grad_norm": 0.5265223024768882,
+      "kl": 24.4734375,
+      "learning_rate": 1.9388679627438486e-05,
+      "loss": 0.9785,
+      "reward": -0.005062588803035339,
+      "reward_std": 0.007155683144557656,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005062588803035339,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 130
+    },
+    {
+      "completion_length": 41.28214565366507,
+      "epoch": 0.20871581795342545,
+      "grad_norm": 0.3295311805017263,
+      "kl": 24.7375,
+      "learning_rate": 1.9292180723175656e-05,
+      "loss": 0.989,
+      "reward": -0.006291940444435884,
+      "reward_std": 0.008898147543002999,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.006291940444435884,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 135
+    },
+    {
+      "completion_length": 57.322324530780314,
+      "epoch": 0.21644603343318194,
+      "grad_norm": 0.30457449118828817,
+      "kl": 23.8953125,
+      "learning_rate": 1.9188890115960967e-05,
+      "loss": 0.9558,
+      "reward": -0.02152316901810991,
+      "reward_std": 0.028220684136340424,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.02152316911124217,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 140
+    },
+    {
+      "completion_length": 55.131253272295,
+      "epoch": 0.22417624891293844,
+      "grad_norm": 9.744918685806544,
+      "kl": 24.059375,
+      "learning_rate": 1.9078883301433488e-05,
+      "loss": 0.9623,
+      "reward": -0.011543530960179993,
+      "reward_std": 0.015249428966203026,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.011543530888875608,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 145
+    },
+    {
+      "completion_length": 41.46696667671203,
+      "epoch": 0.23190646439269494,
+      "grad_norm": 0.43447184518856796,
+      "kl": 24.4078125,
+      "learning_rate": 1.8962240684142923e-05,
+      "loss": 0.9759,
+      "reward": -0.004864498193681755,
+      "reward_std": 0.006879439238022656,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004864498193681755,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 150
+    },
+    {
+      "completion_length": 54.62500326633453,
+      "epoch": 0.23963667987245144,
+      "grad_norm": 0.40864997107181666,
+      "kl": 23.878125,
+      "learning_rate": 1.883904751878156e-05,
+      "loss": 0.9547,
+      "reward": -0.016519928437502072,
+      "reward_std": 0.02108374082057196,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.016519928484068202,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 155
+    },
+    {
+      "completion_length": 39.67232409119606,
+      "epoch": 0.24736689535220793,
+      "grad_norm": 1.4280984501478986,
+      "kl": 24.55625,
+      "learning_rate": 1.8709393847871146e-05,
+      "loss": 0.9824,
+      "reward": -0.011015968853814683,
+      "reward_std": 0.015578932699850157,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.011015968853814683,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 160
+    },
+    {
+      "completion_length": 50.396431642770764,
+      "epoch": 0.25509711083196446,
+      "grad_norm": 2.250106246341844,
+      "kl": 23.9375,
+      "learning_rate": 1.857337443595034e-05,
+      "loss": 0.9575,
+      "reward": -0.011405460795862155,
+      "reward_std": 0.0160778320284237,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.011405460842428283,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 165
+    },
+    {
+      "completion_length": 53.47857511490584,
+      "epoch": 0.26282732631172095,
+      "grad_norm": 0.18024058379756758,
+      "kl": 23.8296875,
+      "learning_rate": 1.8431088700310846e-05,
+      "loss": 0.9531,
+      "reward": -0.0076960978228939325,
+      "reward_std": 0.010883673718687703,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.007696097834535465,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 170
+    },
+    {
+      "completion_length": 28.666966377198698,
+      "epoch": 0.27055754179147745,
+      "grad_norm": 0.1808226277379094,
+      "kl": 24.825,
+      "learning_rate": 1.8282640638332773e-05,
+      "loss": 0.9928,
+      "reward": -0.0072919426711280265,
+      "reward_std": 0.010312364233993777,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0072919426711280265,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 175
+    },
+    {
+      "completion_length": 69.74196844100952,
+      "epoch": 0.27828775727123395,
+      "grad_norm": 0.9161628639155598,
+      "kl": 23.7890625,
+      "learning_rate": 1.8128138751472432e-05,
+      "loss": 0.9512,
+      "reward": -0.030223740078508855,
+      "reward_std": 0.03858855614598724,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.030223740078508855,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 180
+    },
+    {
+      "completion_length": 32.30446657538414,
+      "epoch": 0.28601797275099045,
+      "grad_norm": 0.29658338433610215,
+      "kl": 24.709375,
+      "learning_rate": 1.7967695965958044e-05,
+      "loss": 0.988,
+      "reward": -0.012446127571445231,
+      "reward_std": 0.017601482473719444,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.012446127571445231,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 185
+    },
+    {
+      "completion_length": 30.228573608398438,
+      "epoch": 0.29374818823074694,
+      "grad_norm": 0.3254730765698578,
+      "kl": 24.8359375,
+      "learning_rate": 1.780142955025139e-05,
+      "loss": 0.9932,
+      "reward": -0.00898602957604453,
+      "reward_std": 0.01197805538540706,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.00898602957604453,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 190
+    },
+    {
+      "completion_length": 73.15804052352905,
+      "epoch": 0.30147840371050344,
+      "grad_norm": 0.4643206631442245,
+      "kl": 23.8015625,
+      "learning_rate": 1.7629461029335683e-05,
+      "loss": 0.9521,
+      "reward": -0.014026306665358135,
+      "reward_std": 0.0190565431375731,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.014026306572225877,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 195
+    },
+    {
+      "completion_length": 32.25268087387085,
+      "epoch": 0.30920861919025994,
+      "grad_norm": 3.276210589986656,
+      "kl": 25.0859375,
+      "learning_rate": 1.745191609589231e-05,
+      "loss": 1.0032,
+      "reward": -0.0027048203692515926,
+      "reward_std": 0.003825193697048235,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0027048203692515926,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.30920861919025994,
+      "eval_completion_length": 28.401787757873535,
+      "eval_kl": 25.015625,
+      "eval_loss": 1.002211570739746,
+      "eval_reward": -0.006750390573870391,
+      "eval_reward_std": 0.009546494344249368,
+      "eval_rewards/accuracy_reward": 0.0,
+      "eval_rewards/cosine_scaled_reward": -0.006750390573870391,
+      "eval_rewards/format_reward": 0.0,
+      "eval_rewards/reasoning_steps_reward": 0.0,
+      "eval_runtime": 19.1542,
+      "eval_samples_per_second": 5.169,
+      "eval_steps_per_second": 0.209,
+      "step": 200
+    },
+    {
+      "completion_length": 60.183039855957034,
+      "epoch": 0.31693883467001643,
+      "grad_norm": 4.897646020908246,
+      "kl": 24.5140625,
+      "learning_rate": 1.7268924518431437e-05,
+      "loss": 0.9805,
+      "reward": -0.01982418053667061,
+      "reward_std": 0.025055142166092993,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.01982418053667061,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 205
+    },
+    {
+      "completion_length": 58.75803949385882,
+      "epoch": 0.32466905014977293,
+      "grad_norm": 0.8384849107120312,
+      "kl": 24.6359375,
+      "learning_rate": 1.7080620046443503e-05,
+      "loss": 0.9857,
+      "reward": -0.026825059498571592,
+      "reward_std": 0.03585605600553663,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.026825059312307074,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 210
+    },
+    {
+      "completion_length": 51.62143214643002,
+      "epoch": 0.33239926562952943,
+      "grad_norm": 0.8197769021473577,
+      "kl": 24.45625,
+      "learning_rate": 1.6887140312641036e-05,
+      "loss": 0.9783,
+      "reward": -0.016619097194120514,
+      "reward_std": 0.023502952544595244,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.016619097194120514,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 215
+    },
+    {
+      "completion_length": 57.10357542037964,
+      "epoch": 0.3401294811092859,
+      "grad_norm": 0.7958832727235494,
+      "kl": 24.4234375,
+      "learning_rate": 1.6688626732362192e-05,
+      "loss": 0.9768,
+      "reward": -0.019711116144026163,
+      "reward_std": 0.02642483902600361,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.019711116144026163,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 220
+    },
+    {
+      "completion_length": 59.46964702755213,
+      "epoch": 0.3478596965890424,
+      "grad_norm": 1.417234140942493,
+      "kl": 24.39375,
+      "learning_rate": 1.6485224400209557e-05,
+      "loss": 0.9759,
+      "reward": -0.013211684035978522,
+      "reward_std": 0.018684142745073018,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.013211684035978522,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 225
+    },
+    {
+      "completion_length": 59.63036117553711,
+      "epoch": 0.3555899120687989,
+      "grad_norm": 11.351168718494195,
+      "kl": 24.5234375,
+      "learning_rate": 1.6277081983999742e-05,
+      "loss": 0.9808,
+      "reward": -0.006031500615063123,
+      "reward_std": 0.007947441199212335,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.006031500615063123,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 230
+    },
+    {
+      "completion_length": 49.411610753834246,
+      "epoch": 0.3633201275485554,
+      "grad_norm": 0.2450424456932879,
+      "kl": 24.4359375,
+      "learning_rate": 1.6064351616101318e-05,
+      "loss": 0.9771,
+      "reward": -0.002330746441202791,
+      "reward_std": 0.0030362100929778534,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.002330746441202791,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 235
+    },
+    {
+      "completion_length": 64.94732599258423,
+      "epoch": 0.3710503430283119,
+      "grad_norm": 0.13709329697972883,
+      "kl": 23.971875,
+      "learning_rate": 1.5847188782240473e-05,
+      "loss": 0.9589,
+      "reward": -0.005134024492872413,
+      "reward_std": 0.007109380475594662,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005134024492872413,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 240
+    },
+    {
+      "completion_length": 56.71696834564209,
+      "epoch": 0.3787805585080684,
+      "grad_norm": 0.34364341035671037,
+      "kl": 24.2953125,
+      "learning_rate": 1.562575220785569e-05,
+      "loss": 0.9718,
+      "reward": -0.004507340443524299,
+      "reward_std": 0.006126810604473576,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0045073404202412345,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 245
+    },
+    {
+      "completion_length": 54.89018211364746,
+      "epoch": 0.3865107739878249,
+      "grad_norm": 0.31918456472738405,
+      "kl": 24.278125,
+      "learning_rate": 1.5400203742084508e-05,
+      "loss": 0.9712,
+      "reward": -0.004059947363566607,
+      "reward_std": 0.005513354870345211,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004059947363566607,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 250
+    },
+    {
+      "completion_length": 60.591968631744386,
+      "epoch": 0.3942409894675814,
+      "grad_norm": 0.09960417973828523,
+      "kl": 24.21875,
+      "learning_rate": 1.5170708239467143e-05,
+      "loss": 0.9687,
+      "reward": -0.004857293945678976,
+      "reward_std": 0.006554045081429649,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004857293945678976,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 255
+    },
+    {
+      "completion_length": 51.23661069869995,
+      "epoch": 0.4019712049473379,
+      "grad_norm": 0.26887200107803433,
+      "kl": 24.3703125,
+      "learning_rate": 1.4937433439453465e-05,
+      "loss": 0.9748,
+      "reward": -0.0037404074450023473,
+      "reward_std": 0.005006157659227029,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0037404074450023473,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 260
+    },
+    {
+      "completion_length": 73.15804061889648,
+      "epoch": 0.4097014204270944,
+      "grad_norm": 0.41995946080219004,
+      "kl": 23.8078125,
+      "learning_rate": 1.4700549843801359e-05,
+      "loss": 0.9523,
+      "reward": -0.006090243512880988,
+      "reward_std": 0.008168321024641045,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.006090243501239457,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 265
+    },
+    {
+      "completion_length": 40.091073989868164,
+      "epoch": 0.4174316359068509,
+      "grad_norm": 0.13871083018457628,
+      "kl": 24.6765625,
+      "learning_rate": 1.4460230591956097e-05,
+      "loss": 0.9868,
+      "reward": -0.004075855828705244,
+      "reward_std": 0.005567854072432965,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004075855828705244,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 270
+    },
+    {
+      "completion_length": 70.68839724510909,
+      "epoch": 0.4251618513866074,
+      "grad_norm": 2.5029688664854604,
+      "kl": 23.7,
+      "learning_rate": 1.421665133450184e-05,
+      "loss": 0.9478,
+      "reward": -0.007688843427704484,
+      "reward_std": 0.009824345738056194,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.007688843381138355,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 275
+    },
+    {
+      "completion_length": 70.8294689655304,
+      "epoch": 0.4328920668663639,
+      "grad_norm": 0.7981016773555033,
+      "kl": 23.78125,
+      "learning_rate": 1.3969990104777712e-05,
+      "loss": 0.9508,
+      "reward": -0.01695473511936143,
+      "reward_std": 0.020657122033480847,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.01695473511936143,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 280
+    },
+    {
+      "completion_length": 35.90268077850342,
+      "epoch": 0.4406222823461204,
+      "grad_norm": 0.3832151347916258,
+      "kl": 24.725,
+      "learning_rate": 1.3720427188752306e-05,
+      "loss": 0.9887,
+      "reward": -0.011472697817953303,
+      "reward_std": 0.01592675215797499,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.011472697817953303,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 285
+    },
+    {
+      "completion_length": 57.62857508659363,
+      "epoch": 0.4483524978258769,
+      "grad_norm": 0.19885189538673612,
+      "kl": 24.1484375,
+      "learning_rate": 1.3468144993251735e-05,
+      "loss": 0.9658,
+      "reward": -0.003383369551539772,
+      "reward_std": 0.004771871283338669,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.003383369551539772,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 290
+    },
+    {
+      "completion_length": 67.7410757035017,
+      "epoch": 0.4560827133056334,
+      "grad_norm": 0.39795127039159983,
+      "kl": 23.8921875,
+      "learning_rate": 1.3213327912637563e-05,
+      "loss": 0.9556,
+      "reward": -0.004046605931318936,
+      "reward_std": 0.005629012711131054,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004046605919677404,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 295
+    },
+    {
+      "completion_length": 42.059824275970456,
+      "epoch": 0.4638129287853899,
+      "grad_norm": 0.7635110738781556,
+      "kl": 24.8,
+      "learning_rate": 1.295616219403197e-05,
+      "loss": 0.9916,
+      "reward": -0.004135797415787579,
+      "reward_std": 0.005848900706959625,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004135797415787579,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.4638129287853899,
+      "eval_completion_length": 54.58571767807007,
+      "eval_kl": 25.765625,
+      "eval_loss": 1.032542109489441,
+      "eval_reward": -0.0010169300790039415,
+      "eval_reward_std": 0.00139449925700319,
+      "eval_rewards/accuracy_reward": 0.0,
+      "eval_rewards/cosine_scaled_reward": -0.0010169300790039415,
+      "eval_rewards/format_reward": 0.0,
+      "eval_rewards/reasoning_steps_reward": 0.0,
+      "eval_runtime": 22.4735,
+      "eval_samples_per_second": 4.405,
+      "eval_steps_per_second": 0.178,
+      "step": 300
+    },
+    {
+      "completion_length": 44.41518162488937,
+      "epoch": 0.4715431442651464,
+      "grad_norm": 0.9661396696269151,
+      "kl": 24.990625,
+      "learning_rate": 1.2696835801188816e-05,
+      "loss": 0.9997,
+      "reward": -0.0034492652194785477,
+      "reward_std": 0.004716275511600543,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0034492652194785477,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 305
+    },
+    {
+      "completion_length": 64.66964694261551,
+      "epoch": 0.47927335974490287,
+      "grad_norm": 4.392591196553908,
+      "kl": 34.7390625,
+      "learning_rate": 1.2435538277109919e-05,
+      "loss": 1.3909,
+      "reward": -0.004657760824976265,
+      "reward_std": 0.006355093454780558,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004657760836617797,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 310
+    },
+    {
+      "completion_length": 65.52232588529587,
+      "epoch": 0.48700357522465937,
+      "grad_norm": 1.1190324117378256,
+      "kl": 24.025,
+      "learning_rate": 1.2172460605507126e-05,
+      "loss": 0.961,
+      "reward": -0.010271642655865776,
+      "reward_std": 0.013666120060142362,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.010271642562733519,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 315
+    },
+    {
+      "completion_length": 52.01875347495079,
+      "epoch": 0.49473379070441587,
+      "grad_norm": 0.31844192680864797,
+      "kl": 24.56875,
+      "learning_rate": 1.19077950712113e-05,
+      "loss": 0.9827,
+      "reward": -0.011385823189840582,
+      "reward_std": 0.015236640646251765,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.011385823189840582,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 320
+    },
+    {
+      "completion_length": 58.42857546210289,
+      "epoch": 0.5024640061841724,
+      "grad_norm": 0.41906994568308437,
+      "kl": 24.1828125,
+      "learning_rate": 1.1641735119630373e-05,
+      "loss": 0.967,
+      "reward": -0.009455332184825239,
+      "reward_std": 0.013371859212202252,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.009455332184825239,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 325
+    },
+    {
+      "completion_length": 48.32500350177288,
+      "epoch": 0.5101942216639289,
+      "grad_norm": 0.15370140997531662,
+      "kl": 24.446875,
+      "learning_rate": 1.137447521535908e-05,
+      "loss": 0.9778,
+      "reward": -0.003301967513988302,
+      "reward_std": 0.004669687197258554,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.003301967513988302,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 330
+    },
+    {
+      "completion_length": 69.40357585400344,
+      "epoch": 0.5179244371436854,
+      "grad_norm": 0.2903211802770792,
+      "kl": 23.709375,
+      "learning_rate": 1.110621070004378e-05,
+      "loss": 0.9482,
+      "reward": -0.0057826179621066846,
+      "reward_std": 0.008032065768097141,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005782618008672813,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 335
+    },
+    {
+      "completion_length": 40.13571699261665,
+      "epoch": 0.5256546526234419,
+      "grad_norm": 0.0914726355478226,
+      "kl": 24.45625,
+      "learning_rate": 1.0837137649606241e-05,
+      "loss": 0.978,
+      "reward": -0.0011966714072912943,
+      "reward_std": 0.0016399607283785755,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0011966714072912943,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 340
+    },
+    {
+      "completion_length": 75.69375510960818,
+      "epoch": 0.5333848681031984,
+      "grad_norm": 0.11627122046037379,
+      "kl": 23.6296875,
+      "learning_rate": 1.0567452730930743e-05,
+      "loss": 0.9452,
+      "reward": -0.0014592139999309417,
+      "reward_std": 0.002016079255188963,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0014592139941101757,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 345
+    },
+    {
+      "completion_length": 54.79911108016968,
+      "epoch": 0.5411150835829549,
+      "grad_norm": 0.1561203540032164,
+      "kl": 24.1921875,
+      "learning_rate": 1.0297353058119209e-05,
+      "loss": 0.9674,
+      "reward": -0.0011425426000443828,
+      "reward_std": 0.0016062345731212347,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0011425426000443828,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 350
+    },
+    {
+      "completion_length": 45.740181751549244,
+      "epoch": 0.5488452990627114,
+      "grad_norm": 0.18815236709612004,
+      "kl": 24.3046875,
+      "learning_rate": 1.0027036048419514e-05,
+      "loss": 0.972,
+      "reward": -0.0014331862192591416,
+      "reward_std": 0.002026831388690198,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0014331862192591416,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 355
+    },
+    {
+      "completion_length": 79.47232694625855,
+      "epoch": 0.5565755145424679,
+      "grad_norm": 0.8479026792877249,
+      "kl": 23.4390625,
+      "learning_rate": 9.756699277932196e-06,
+      "loss": 0.9374,
+      "reward": -0.0020212634304677124,
+      "reward_std": 0.0027920879992834103,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0020212634333780953,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 360
+    },
+    {
+      "completion_length": 60.38214693069458,
+      "epoch": 0.5643057300222244,
+      "grad_norm": 0.11021078270039414,
+      "kl": 24.025,
+      "learning_rate": 9.486540337201046e-06,
+      "loss": 0.9611,
+      "reward": -0.0012649027251427469,
+      "reward_std": 0.0017696052445899113,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0012649027251427469,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 365
+    },
+    {
+      "completion_length": 35.70893115997315,
+      "epoch": 0.5720359455019809,
+      "grad_norm": 0.14685497696691516,
+      "kl": 24.5890625,
+      "learning_rate": 9.216756686793163e-06,
+      "loss": 0.9834,
+      "reward": -0.0007313355393591791,
+      "reward_std": 0.0010342646257754495,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0007313355393591791,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 370
+    },
+    {
+      "completion_length": 42.10268154144287,
+      "epoch": 0.5797661609817374,
+      "grad_norm": 0.09976156130247066,
+      "kl": 24.515625,
+      "learning_rate": 8.94754551297402e-06,
+      "loss": 0.9803,
+      "reward": -0.0007080816060238248,
+      "reward_std": 0.0009251405505210641,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0007080815943822927,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 375
+    },
+    {
+      "completion_length": 69.32857606261969,
+      "epoch": 0.5874963764614939,
+      "grad_norm": 0.7754555180333265,
+      "kl": 23.7734375,
+      "learning_rate": 8.67910358358298e-06,
+      "loss": 0.9511,
+      "reward": -0.002085479956159819,
+      "reward_std": 0.0028004977468595984,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.002085479956159819,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 380
+    },
+    {
+      "completion_length": 76.5732192993164,
+      "epoch": 0.5952265919412504,
+      "grad_norm": 0.5202858180308233,
+      "kl": 23.371875,
+      "learning_rate": 8.411627104214675e-06,
+      "loss": 0.935,
+      "reward": -0.0017017564326295086,
+      "reward_std": 0.002331838384964158,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0017017564311743172,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 385
+    },
+    {
+      "completion_length": 62.300896883010864,
+      "epoch": 0.6029568074210069,
+      "grad_norm": 0.08412830920156626,
+      "kl": 23.98125,
+      "learning_rate": 8.145311574811325e-06,
+      "loss": 0.9592,
+      "reward": -0.0019122766454344742,
+      "reward_std": 0.00268942869131763,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0019122766454344742,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 390
+    },
+    {
+      "completion_length": 45.8089316368103,
+      "epoch": 0.6106870229007634,
+      "grad_norm": 0.10752134863872875,
+      "kl": 24.4234375,
+      "learning_rate": 7.880351646770824e-06,
+      "loss": 0.9765,
+      "reward": -0.000848563711724637,
+      "reward_std": 0.0011321354522124238,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.000848563711724637,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 395
+    },
+    {
+      "completion_length": 47.3651818215847,
+      "epoch": 0.6184172383805199,
+      "grad_norm": 0.5988906531284993,
+      "kl": 24.334375,
+      "learning_rate": 7.616940980675004e-06,
+      "loss": 0.973,
+      "reward": -0.0015042206411578717,
+      "reward_std": 0.0021272892814579334,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0015042206411578717,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.6184172383805199,
+      "eval_completion_length": 51.26339612901211,
+      "eval_kl": 24.21875,
+      "eval_loss": 0.9700301289558411,
+      "eval_reward": -0.001125097114993423,
+      "eval_reward_std": 0.0015911276817632825,
+      "eval_rewards/accuracy_reward": 0.0,
+      "eval_rewards/cosine_scaled_reward": -0.0011250970858895926,
+      "eval_rewards/format_reward": 0.0,
+      "eval_rewards/reasoning_steps_reward": 0.0,
+      "eval_runtime": 22.626,
+      "eval_samples_per_second": 4.376,
+      "eval_steps_per_second": 0.177,
+      "step": 400
+    },
+    {
+      "completion_length": 86.76786234378815,
+      "epoch": 0.6261474538602764,
+      "grad_norm": 1.1818892245329233,
+      "kl": 23.203125,
+      "learning_rate": 7.355272104742132e-06,
+      "loss": 0.9282,
+      "reward": -0.0026306660576764784,
+      "reward_std": 0.003647236472497184,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0026306660576764784,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 405
+    },
+    {
+      "completion_length": 62.026789891719815,
+      "epoch": 0.6338776693400329,
+      "grad_norm": 0.2836127578512707,
+      "kl": 23.709375,
+      "learning_rate": 7.095536274107046e-06,
+      "loss": 0.9482,
+      "reward": -0.002841473452235732,
+      "reward_std": 0.004013956811013486,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.002841473452235732,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 410
+    },
+    {
+      "completion_length": 46.67500334680081,
+      "epoch": 0.6416078848197894,
+      "grad_norm": 0.26356911850515,
+      "kl": 24.3359375,
+      "learning_rate": 6.837923331031761e-06,
+      "loss": 0.9733,
+      "reward": -0.001941360929671987,
+      "reward_std": 0.0027454989251992855,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.001941360929671987,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 415
+    },
+    {
+      "completion_length": 37.07410943508148,
+      "epoch": 0.6493381002995459,
+      "grad_norm": 0.2805168584113563,
+      "kl": 24.54375,
+      "learning_rate": 6.58262156614881e-06,
+      "loss": 0.9813,
+      "reward": -0.0028216561851877485,
+      "reward_std": 0.003953768716746709,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0028216561851877485,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 420
+    },
+    {
+      "completion_length": 54.22589612603188,
+      "epoch": 0.6570683157793024,
+      "grad_norm": 0.913450851127483,
+      "kl": 23.8859375,
+      "learning_rate": 6.3298175808386284e-06,
+      "loss": 0.9553,
+      "reward": -0.007596620212734706,
+      "reward_std": 0.009845524301908881,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.007596620212734706,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 425
+    },
+    {
+      "completion_length": 48.00268133282661,
+      "epoch": 0.6647985312590589,
+      "grad_norm": 0.4604626281806278,
+      "kl": 23.9265625,
+      "learning_rate": 6.079696150841634e-06,
+      "loss": 0.9569,
+      "reward": -0.012062773975776509,
+      "reward_std": 0.016299182930379175,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.012062773999059574,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 430
+    },
+    {
+      "completion_length": 46.86160971969366,
+      "epoch": 0.6725287467388154,
+      "grad_norm": 0.7252971846570313,
+      "kl": 24.1859375,
+      "learning_rate": 5.832440091204698e-06,
+      "loss": 0.9676,
+      "reward": -0.012686770772155142,
+      "reward_std": 0.017015842359229795,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.012686770772155142,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 435
+    },
+    {
+      "completion_length": 47.819645977020265,
+      "epoch": 0.6802589622185718,
+      "grad_norm": 0.2805343679934075,
+      "kl": 24.2859375,
+      "learning_rate": 5.588230122660672e-06,
+      "loss": 0.9712,
+      "reward": -0.012813755803776417,
+      "reward_std": 0.017698440483681566,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.012813755803776417,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 440
+    },
+    {
+      "completion_length": 51.863396406173706,
+      "epoch": 0.6879891776983283,
+      "grad_norm": 0.309711353898578,
+      "kl": 24.31875,
+      "learning_rate": 5.347244739538677e-06,
+      "loss": 0.9726,
+      "reward": -0.021797565624001435,
+      "reward_std": 0.027237284120928962,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.021797565717133694,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 445
+    },
+    {
+      "completion_length": 64.80982577204705,
+      "epoch": 0.6957193931780848,
+      "grad_norm": 1.2162712083436764,
+      "kl": 24.14375,
+      "learning_rate": 5.109660079301668e-06,
+      "loss": 0.9659,
+      "reward": -0.02888913472050376,
+      "reward_std": 0.04085540586487468,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.028889134627371506,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 450
+    },
+    {
+      "completion_length": 59.602682419121265,
+      "epoch": 0.7034496086578413,
+      "grad_norm": 0.3933355097706156,
+      "kl": 24.1421875,
+      "learning_rate": 4.875649793806655e-06,
+      "loss": 0.9656,
+      "reward": -0.026990144280716777,
+      "reward_std": 0.03466883820692601,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.026990144187584518,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 455
+    },
+    {
+      "completion_length": 31.516073471307756,
+      "epoch": 0.7111798241375978,
+      "grad_norm": 0.8420503487470977,
+      "kl": 24.99375,
+      "learning_rate": 4.64538492238166e-06,
+      "loss": 0.9995,
+      "reward": -0.007723358142743564,
+      "reward_std": 0.010892791659921386,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.007723358142743564,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 460
+    },
+    {
+      "completion_length": 41.60357391834259,
+      "epoch": 0.7189100396173543,
+      "grad_norm": 0.8662013381034004,
+      "kl": 24.68125,
+      "learning_rate": 4.4190337668121964e-06,
+      "loss": 0.9872,
+      "reward": -0.009986836141189403,
+      "reward_std": 0.013905720249748121,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.009986836187755532,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 465
+    },
+    {
+      "completion_length": 50.748217730224134,
+      "epoch": 0.7266402550971108,
+      "grad_norm": 0.28970671122261415,
+      "kl": 24.2953125,
+      "learning_rate": 4.196761768328599e-06,
+      "loss": 0.9714,
+      "reward": -0.014916278285645745,
+      "reward_std": 0.021028667081813522,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.014916278285645745,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 470
+    },
+    {
+      "completion_length": 68.74196868389845,
+      "epoch": 0.7343704705768673,
+      "grad_norm": 0.918329541521939,
+      "kl": 24.0203125,
+      "learning_rate": 3.978731386684206e-06,
+      "loss": 0.9612,
+      "reward": -0.016641861237076228,
+      "reward_std": 0.022046213136218285,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.016641861237076228,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 475
+    },
+    {
+      "completion_length": 53.72321764677763,
+      "epoch": 0.7421006860566238,
+      "grad_norm": 78.83378790639892,
+      "kl": 25.0453125,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 1.0021,
+      "reward": -0.009450171658966245,
+      "reward_std": 0.012123302213909249,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.00945017168224931,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 480
+    },
+    {
+      "completion_length": 51.16786060333252,
+      "epoch": 0.7498309015363803,
+      "grad_norm": 0.33308461385913096,
+      "kl": 24.4515625,
+      "learning_rate": 3.5560296953512296e-06,
+      "loss": 0.9779,
+      "reward": -0.004581006449978986,
+      "reward_std": 0.006231808224913493,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004581006496545114,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 485
+    },
+    {
+      "completion_length": 55.30178937315941,
+      "epoch": 0.7575611170161368,
+      "grad_norm": 305.81882755074383,
+      "kl": 27.3046875,
+      "learning_rate": 3.3516673405151546e-06,
+      "loss": 1.0928,
+      "reward": -0.003802679693740174,
+      "reward_std": 0.005377801274141803,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0038026797170232383,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 490
+    },
+    {
+      "completion_length": 63.23661144077778,
+      "epoch": 0.7652913324958933,
+      "grad_norm": 0.5787231505930388,
+      "kl": 25.1375,
+      "learning_rate": 3.1521642864065905e-06,
+      "loss": 1.0056,
+      "reward": -0.004590116689792101,
+      "reward_std": 0.006453327897092543,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004590116689792101,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 495
+    },
+    {
+      "completion_length": 46.05268131196499,
+      "epoch": 0.7730215479756498,
+      "grad_norm": 1.435679878343505,
+      "kl": 24.3078125,
+      "learning_rate": 2.957666350839663e-06,
+      "loss": 0.9722,
+      "reward": -0.0046006116961741835,
+      "reward_std": 0.006438373773643491,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0046006116961741835,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.7730215479756498,
+      "eval_completion_length": 40.11160922050476,
+      "eval_kl": 24.578125,
+      "eval_loss": 0.985271155834198,
+      "eval_reward": -0.006326919745333726,
+      "eval_reward_std": 0.008947615635406692,
+      "eval_rewards/accuracy_reward": 0.0,
+      "eval_rewards/cosine_scaled_reward": -0.006326919745333726,
+      "eval_rewards/format_reward": 0.0,
+      "eval_rewards/reasoning_steps_reward": 0.0,
+      "eval_runtime": 23.2741,
+      "eval_samples_per_second": 4.254,
+      "eval_steps_per_second": 0.172,
+      "step": 500
+    },
+    {
+      "completion_length": 36.7062520518899,
+      "epoch": 0.7807517634554063,
+      "grad_norm": 0.2761808288916843,
+      "kl": 24.5265625,
+      "learning_rate": 2.768315693361474e-06,
+      "loss": 0.981,
+      "reward": -0.005745776838499239,
+      "reward_std": 0.008125755542479851,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005745776745366982,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 505
+    },
+    {
+      "completion_length": 44.904467339813706,
+      "epoch": 0.7884819789351628,
+      "grad_norm": 1.3465156523462427,
+      "kl": 24.49375,
+      "learning_rate": 2.5842507113469307e-06,
+      "loss": 0.98,
+      "reward": -0.005905044664916659,
+      "reward_std": 0.008296428343896878,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005905044664916659,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 510
+    },
+    {
+      "completion_length": 42.89464537650347,
+      "epoch": 0.7962121944149193,
+      "grad_norm": 0.6423998335451256,
+      "kl": 24.3765625,
+      "learning_rate": 2.405605938843416e-06,
+      "loss": 0.975,
+      "reward": -0.005736959281955478,
+      "reward_std": 0.008099251179783096,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005736959281955478,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 515
+    },
+    {
+      "completion_length": 48.71964579820633,
+      "epoch": 0.8039424098946758,
+      "grad_norm": 0.9616888653441464,
+      "kl": 24.0609375,
+      "learning_rate": 2.2325119482391466e-06,
+      "loss": 0.9623,
+      "reward": -0.005028559759688278,
+      "reward_std": 0.007111457411508581,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005028559713122149,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 520
+    },
+    {
+      "completion_length": 39.69464554786682,
+      "epoch": 0.8116726253744323,
+      "grad_norm": 10.30615008769208,
+      "kl": 24.7109375,
+      "learning_rate": 2.065095254827133e-06,
+      "loss": 0.9885,
+      "reward": -0.0030483946065032797,
+      "reward_std": 0.004290749793926807,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0030483946094136626,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 525
+    },
+    {
+      "completion_length": 44.354466861486436,
+      "epoch": 0.8194028408541888,
+      "grad_norm": 0.615670728640926,
+      "kl": 24.409375,
+      "learning_rate": 1.9034782243345074e-06,
+      "loss": 0.9759,
+      "reward": -0.0030309244835677873,
+      "reward_std": 0.004090318087315836,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0030309244893885535,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 530
+    },
+    {
+      "completion_length": 46.14911011010408,
+      "epoch": 0.8271330563339453,
+      "grad_norm": 0.7617698847734057,
+      "kl": 24.4140625,
+      "learning_rate": 1.7477789834847835e-06,
+      "loss": 0.9761,
+      "reward": -0.004230587323957202,
+      "reward_std": 0.005964093679218863,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.00423058731231567,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 535
+    },
+    {
+      "completion_length": 39.64196678251028,
+      "epoch": 0.8348632718137018,
+      "grad_norm": 1.156878402277619,
+      "kl": 24.2203125,
+      "learning_rate": 1.5981113336584041e-06,
+      "loss": 0.9687,
+      "reward": -0.003543670538948618,
+      "reward_std": 0.004806647020114951,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.003543670538948618,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 540
+    },
+    {
+      "completion_length": 46.57232457995415,
+      "epoch": 0.8425934872934583,
+      "grad_norm": 0.4199458638335275,
+      "kl": 24.2046875,
+      "learning_rate": 1.4545846677147446e-06,
+      "loss": 0.9682,
+      "reward": -0.005338836487372589,
+      "reward_std": 0.007550142756585387,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005338836487372589,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 545
+    },
+    {
+      "completion_length": 49.66607456803322,
+      "epoch": 0.8503237027732148,
+      "grad_norm": 4.461365741300084,
+      "kl": 24.1,
+      "learning_rate": 1.3173038900362977e-06,
+      "loss": 0.964,
+      "reward": -0.004732385008082929,
+      "reward_std": 0.00659514378021413,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004732385008082929,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 550
+    },
+    {
+      "completion_length": 45.15446722209454,
+      "epoch": 0.8580539182529713,
+      "grad_norm": 0.5986641101244399,
+      "kl": 24.3421875,
+      "learning_rate": 1.1863693398535115e-06,
+      "loss": 0.9735,
+      "reward": -0.0029984577427512706,
+      "reward_std": 0.004211953165122395,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0029984577427512706,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 555
+    },
+    {
+      "completion_length": 45.605360022187234,
+      "epoch": 0.8657841337327278,
+      "grad_norm": 0.534747510415695,
+      "kl": 24.2609375,
+      "learning_rate": 1.0618767179063416e-06,
+      "loss": 0.9702,
+      "reward": -0.0036516298819336155,
+      "reward_std": 0.005164184488143065,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.003651629893575148,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 560
+    },
+    {
+      "completion_length": 42.566074170172215,
+      "epoch": 0.8735143492124843,
+      "grad_norm": 0.6059644840217457,
+      "kl": 24.44375,
+      "learning_rate": 9.439170164960765e-07,
+      "loss": 0.9778,
+      "reward": -0.0029312313739865204,
+      "reward_std": 0.004098246799730987,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.002931231374714116,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 565
+    },
+    {
+      "completion_length": 53.61071793586016,
+      "epoch": 0.8812445646922408,
+      "grad_norm": 1.6865381605569074,
+      "kl": 23.9640625,
+      "learning_rate": 8.325764529785851e-07,
+      "loss": 0.9585,
+      "reward": -0.004873991443924553,
+      "reward_std": 0.006856209293836457,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004873991455566084,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 570
+    },
+    {
+      "completion_length": 48.17232428640127,
+      "epoch": 0.8889747801719973,
+      "grad_norm": 0.793839960135767,
+      "kl": 24.1234375,
+      "learning_rate": 7.279364067476247e-07,
+      "loss": 0.9646,
+      "reward": -0.004261988118764748,
+      "reward_std": 0.005693929277595089,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004261988124585514,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 575
+    },
+    {
+      "completion_length": 65.91875450015068,
+      "epoch": 0.8967049956517538,
+      "grad_norm": 0.36075442660566825,
+      "kl": 23.728125,
+      "learning_rate": 6.300733597542086e-07,
+      "loss": 0.949,
+      "reward": -0.004888249606960926,
+      "reward_std": 0.006862056562249563,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004888249606960926,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 580
+    },
+    {
+      "completion_length": 59.56786108762026,
+      "epoch": 0.9044352111315103,
+      "grad_norm": 0.470810200110092,
+      "kl": 23.8546875,
+      "learning_rate": 5.390588406055497e-07,
+      "loss": 0.9541,
+      "reward": -0.004270391982012711,
+      "reward_std": 0.006030588216785304,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004270391982012711,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 585
+    },
+    {
+      "completion_length": 49.16071737408638,
+      "epoch": 0.9121654266112668,
+      "grad_norm": 0.30166398904682923,
+      "kl": 24.2296875,
+      "learning_rate": 4.549593722844492e-07,
+      "loss": 0.969,
+      "reward": -0.0038439762124880873,
+      "reward_std": 0.005080879169804575,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0038439761659219583,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 590
+    },
+    {
+      "completion_length": 41.13482419103384,
+      "epoch": 0.9198956420910233,
+      "grad_norm": 2.14891764621043,
+      "kl": 24.615625,
+      "learning_rate": 3.77836423527278e-07,
+      "loss": 0.9848,
+      "reward": -0.0033783776602923864,
+      "reward_std": 0.004549032651979701,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0033783776602923864,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 595
+    },
+    {
+      "completion_length": 34.614287830889225,
+      "epoch": 0.9276258575707798,
+      "grad_norm": 0.5737064816443503,
+      "kl": 24.590625,
+      "learning_rate": 3.0774636389618196e-07,
+      "loss": 0.9838,
+      "reward": -0.002332200028695297,
+      "reward_std": 0.003298228911899059,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.002332200028695297,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.9276258575707798,
+      "eval_completion_length": 32.74821615219116,
+      "eval_kl": 24.890625,
+      "eval_loss": 0.9939680695533752,
+      "eval_reward": -0.0007343304278037976,
+      "eval_reward_std": 0.0009667417034506798,
+      "eval_rewards/accuracy_reward": 0.0,
+      "eval_rewards/cosine_scaled_reward": -0.0007343304278037976,
+      "eval_rewards/format_reward": 0.0,
+      "eval_rewards/reasoning_steps_reward": 0.0,
+      "eval_runtime": 21.7904,
+      "eval_samples_per_second": 4.543,
+      "eval_steps_per_second": 0.184,
+      "step": 600
+    },
+    {
+      "completion_length": 31.262502002716065,
+      "epoch": 0.9353560730505363,
+      "grad_norm": 2.2163447857886047,
+      "kl": 24.6203125,
+      "learning_rate": 2.44740422578269e-07,
+      "loss": 0.9847,
+      "reward": -0.0034189186366347713,
+      "reward_std": 0.004835081120654649,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0034189186366347713,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 605
+    },
+    {
+      "completion_length": 49.31071757972241,
+      "epoch": 0.9430862885302927,
+      "grad_norm": 0.6229861448868668,
+      "kl": 24.153125,
+      "learning_rate": 1.8886465094192895e-07,
+      "loss": 0.966,
+      "reward": -0.003229186121393468,
+      "reward_std": 0.004545925347724733,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.003229186121393468,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 610
+    },
+    {
+      "completion_length": 35.64910935461521,
+      "epoch": 0.9508165040100492,
+      "grad_norm": 3.705591212787074,
+      "kl": 24.5125,
+      "learning_rate": 1.401598888776523e-07,
+      "loss": 0.9804,
+      "reward": -0.004258286483764096,
+      "reward_std": 0.005848287543631159,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004258286530330224,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 615
+    },
+    {
+      "completion_length": 56.283039236068724,
+      "epoch": 0.9585467194898057,
+      "grad_norm": 1.353088289442673,
+      "kl": 24.2328125,
+      "learning_rate": 9.866173494794462e-08,
+      "loss": 0.9693,
+      "reward": -0.005120481132257737,
+      "reward_std": 0.006864398748734857,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.005120481178823866,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 620
+    },
+    {
+      "completion_length": 38.977681183815,
+      "epoch": 0.9662769349695622,
+      "grad_norm": 0.5783254077528687,
+      "kl": 24.5234375,
+      "learning_rate": 6.440052036815081e-08,
+      "loss": 0.9807,
+      "reward": -0.0028176053295908333,
+      "reward_std": 0.003978943701770277,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0028176053295908333,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 625
+    },
+    {
+      "completion_length": 42.6562525421381,
+      "epoch": 0.9740071504493187,
+      "grad_norm": 0.8232559607607927,
+      "kl": 24.38125,
+      "learning_rate": 3.7401286837214224e-08,
+      "loss": 0.9753,
+      "reward": -0.002573585319900218,
+      "reward_std": 0.0034814744626572745,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.002573585325720984,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 630
+    },
+    {
+      "completion_length": 46.04732434153557,
+      "epoch": 0.9817373659290752,
+      "grad_norm": 0.46809429712914535,
+      "kl": 24.3265625,
+      "learning_rate": 1.7683768234568745e-08,
+      "loss": 0.9731,
+      "reward": -0.0030760798572949,
+      "reward_std": 0.004335028942911379,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0030760798922194964,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 635
+    },
+    {
+      "completion_length": 42.86875238418579,
+      "epoch": 0.9894675814088317,
+      "grad_norm": 0.6530501980892132,
+      "kl": 24.375,
+      "learning_rate": 5.262376196544239e-09,
+      "loss": 0.9751,
+      "reward": -0.004086044424411739,
+      "reward_std": 0.005763764342600552,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.004086044424411739,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 640
+    },
+    {
+      "completion_length": 38.43035954236984,
+      "epoch": 0.9971977968885882,
+      "grad_norm": 0.5860528397565183,
+      "kl": 24.2625,
+      "learning_rate": 1.461895828280824e-10,
+      "loss": 0.9703,
+      "reward": -0.0041267017044091855,
+      "reward_std": 0.0050602933844260175,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0041267017044091855,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 645
+    },
+    {
+      "completion_length": 42.96875283122063,
+      "epoch": 0.9987438399845395,
+      "kl": 24.4140625,
+      "reward": -0.0019939331905334257,
+      "reward_std": 0.0028198473701195326,
+      "rewards/accuracy_reward": 0.0,
+      "rewards/cosine_scaled_reward": -0.0019939331905334257,
+      "rewards/format_reward": 0.0,
+      "rewards/reasoning_steps_reward": 0.0,
+      "step": 646,
+      "total_flos": 0.0,
+      "train_loss": 0.9507393147195028,
+      "train_runtime": 22343.714,
+      "train_samples_per_second": 3.242,
+      "train_steps_per_second": 0.029
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 646,
+  "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": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61f3674824eab11c920a307a67488f2e5e97863445d23083c0d34ae7fe211c7c
+size 7544