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/Qwen2.5-0.5B-Instruct
+library_name: transformers
+model_name: Freeway-GRPO-Qwen-2.5-0.5B-Instruct
+tags:
+- generated_from_trainer
+- trl
+- grpo
+licence: license
+---
+
+# Model Card for Freeway-GRPO-Qwen-2.5-0.5B-Instruct
+
+This model is a fine-tuned version of [Qwen/Qwen2.5-0.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-0.5B-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="wenyl/Freeway-GRPO-Qwen-2.5-0.5B-Instruct", 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/wenyl22/huggingface/runs/azrdhm61) 
+
+
+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: 0.15.2
+- Transformers: 4.49.0
+- Pytorch: 2.5.1
+- Datasets: 3.4.0
+- Tokenizers: 0.21.1
+
+## 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": 0.004944845609998083,
+    "train_runtime": 5152.2436,
+    "train_samples": 200080,
+    "train_samples_per_second": 3.18,
+    "train_steps_per_second": 0.05
+}

config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "Qwen/Qwen2.5-0.5B-Instruct",
+  "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": 32768,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "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"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e26af9bb4d4d9256f7d456271e45e44685c8b966b59e4d580c5d08fd3170ea0
+size 988097824

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 +1,208 @@
+{
+  "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|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "total_flos": 0.0,
+    "train_loss": 0.004944845609998083,
+    "train_runtime": 5152.2436,
+    "train_samples": 200080,
+    "train_samples_per_second": 3.18,
+    "train_steps_per_second": 0.05
+}

trainer_state.json

@@ -0,0 +1,774 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.04094362255097961,
+  "eval_steps": 64,
+  "global_step": 256,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "completion_length": 52.6875,
+      "epoch": 0.0001599360255897641,
+      "grad_norm": 6.5625,
+      "kl": 0.0,
+      "learning_rate": 1.1538461538461539e-07,
+      "loss": -0.0,
+      "reward": 0.359375,
+      "reward_std": 0.3756504710763693,
+      "rewards/accuracy_reward": 0.171875,
+      "rewards/format_reward": 0.1875,
+      "step": 1
+    },
+    {
+      "completion_length": 49.2109375,
+      "epoch": 0.0007996801279488205,
+      "grad_norm": 5.3125,
+      "kl": 0.0012457009133868269,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 0.0,
+      "reward": 0.48828125,
+      "reward_std": 0.42536891577765346,
+      "rewards/accuracy_reward": 0.28515625,
+      "rewards/format_reward": 0.203125,
+      "step": 5
+    },
+    {
+      "completion_length": 53.75625,
+      "epoch": 0.001599360255897641,
+      "grad_norm": 7.15625,
+      "kl": 0.0015386140597911436,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 0.0001,
+      "reward": 0.46875,
+      "reward_std": 0.4507805652916431,
+      "rewards/accuracy_reward": 0.228125,
+      "rewards/format_reward": 0.240625,
+      "step": 10
+    },
+    {
+      "completion_length": 56.421875,
+      "epoch": 0.0023990403838464614,
+      "grad_norm": 6.90625,
+      "kl": 0.0025526953544613208,
+      "learning_rate": 1.7307692307692306e-06,
+      "loss": 0.0001,
+      "reward": 0.51875,
+      "reward_std": 0.47729706913232806,
+      "rewards/accuracy_reward": 0.303125,
+      "rewards/format_reward": 0.215625,
+      "step": 15
+    },
+    {
+      "completion_length": 50.725,
+      "epoch": 0.003198720511795282,
+      "grad_norm": 5.4375,
+      "kl": 0.007305113899565186,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 0.0003,
+      "reward": 0.584375,
+      "reward_std": 0.4640388172119856,
+      "rewards/accuracy_reward": 0.3,
+      "rewards/format_reward": 0.284375,
+      "step": 20
+    },
+    {
+      "completion_length": 48.2625,
+      "epoch": 0.003998400639744102,
+      "grad_norm": 7.59375,
+      "kl": 0.021931850849068724,
+      "learning_rate": 2.884615384615385e-06,
+      "loss": 0.0009,
+      "reward": 0.709375,
+      "reward_std": 0.49055532105267047,
+      "rewards/accuracy_reward": 0.35,
+      "rewards/format_reward": 0.359375,
+      "step": 25
+    },
+    {
+      "completion_length": 46.578125,
+      "epoch": 0.004798080767692923,
+      "grad_norm": 5.375,
+      "kl": 0.08235460844589397,
+      "learning_rate": 2.997761705224242e-06,
+      "loss": 0.0033,
+      "reward": 0.840625,
+      "reward_std": 0.5524271633476019,
+      "rewards/accuracy_reward": 0.384375,
+      "rewards/format_reward": 0.45625,
+      "step": 30
+    },
+    {
+      "completion_length": 50.603125,
+      "epoch": 0.005597760895641743,
+      "grad_norm": 5.625,
+      "kl": 0.12322683255188167,
+      "learning_rate": 2.9886800800368023e-06,
+      "loss": 0.0049,
+      "reward": 1.01875,
+      "reward_std": 0.47729706913232806,
+      "rewards/accuracy_reward": 0.39375,
+      "rewards/format_reward": 0.625,
+      "step": 35
+    },
+    {
+      "completion_length": 51.74375,
+      "epoch": 0.006397441023590564,
+      "grad_norm": 5.25,
+      "kl": 0.12255409228382633,
+      "learning_rate": 2.972657541113338e-06,
+      "loss": 0.0049,
+      "reward": 1.05,
+      "reward_std": 0.4596193999052048,
+      "rewards/accuracy_reward": 0.403125,
+      "rewards/format_reward": 0.646875,
+      "step": 40
+    },
+    {
+      "completion_length": 51.15,
+      "epoch": 0.007197121151539384,
+      "grad_norm": 4.34375,
+      "kl": 0.11809529251186177,
+      "learning_rate": 2.949768792926617e-06,
+      "loss": 0.0047,
+      "reward": 1.1625,
+      "reward_std": 0.5303300768136978,
+      "rewards/accuracy_reward": 0.45,
+      "rewards/format_reward": 0.7125,
+      "step": 45
+    },
+    {
+      "completion_length": 52.084375,
+      "epoch": 0.007996801279488205,
+      "grad_norm": 5.09375,
+      "kl": 0.14425031405407934,
+      "learning_rate": 2.9201205533865656e-06,
+      "loss": 0.0058,
+      "reward": 1.096875,
+      "reward_std": 0.5170718248933553,
+      "rewards/accuracy_reward": 0.41875,
+      "rewards/format_reward": 0.678125,
+      "step": 50
+    },
+    {
+      "completion_length": 49.265625,
+      "epoch": 0.008796481407437025,
+      "grad_norm": 5.25,
+      "kl": 0.12807998866774142,
+      "learning_rate": 2.8838510562721076e-06,
+      "loss": 0.0051,
+      "reward": 1.08125,
+      "reward_std": 0.5038135729730129,
+      "rewards/accuracy_reward": 0.396875,
+      "rewards/format_reward": 0.684375,
+      "step": 55
+    },
+    {
+      "completion_length": 53.115625,
+      "epoch": 0.009596161535385846,
+      "grad_norm": 5.09375,
+      "kl": 0.12884393450804055,
+      "learning_rate": 2.8411294067214768e-06,
+      "loss": 0.0052,
+      "reward": 1.125,
+      "reward_std": 0.5391689114272594,
+      "rewards/accuracy_reward": 0.45,
+      "rewards/format_reward": 0.675,
+      "step": 60
+    },
+    {
+      "epoch": 0.010235905637744902,
+      "eval_completion_length": 51.59410112359551,
+      "eval_kl": 0.12826356122249297,
+      "eval_loss": 0.004409829154610634,
+      "eval_reward": 1.0603932584269662,
+      "eval_reward_std": 0.5144400355521213,
+      "eval_rewards/accuracy_reward": 0.43820224719101125,
+      "eval_rewards/format_reward": 0.6221910112359551,
+      "eval_runtime": 19.3936,
+      "eval_samples_per_second": 5.156,
+      "eval_steps_per_second": 1.289,
+      "step": 64
+    },
+    {
+      "completion_length": 48.90625,
+      "epoch": 0.010395841663334666,
+      "grad_norm": 4.09375,
+      "kl": 0.1414291060063988,
+      "learning_rate": 2.792154792786e-06,
+      "loss": 0.0055,
+      "reward": 1.03125,
+      "reward_std": 0.4419417306780815,
+      "rewards/accuracy_reward": 0.40625,
+      "rewards/format_reward": 0.625,
+      "step": 65
+    },
+    {
+      "completion_length": 49.440625,
+      "epoch": 0.011195521791283487,
+      "grad_norm": 5.25,
+      "kl": 0.11723160503897816,
+      "learning_rate": 2.737155556723453e-06,
+      "loss": 0.0047,
+      "reward": 1.053125,
+      "reward_std": 0.49055532105267047,
+      "rewards/accuracy_reward": 0.38125,
+      "rewards/format_reward": 0.671875,
+      "step": 70
+    },
+    {
+      "completion_length": 51.290625,
+      "epoch": 0.011995201919232307,
+      "grad_norm": 4.53125,
+      "kl": 0.12841552164172754,
+      "learning_rate": 2.676388130361047e-06,
+      "loss": 0.0051,
+      "reward": 1.084375,
+      "reward_std": 0.5170718248933553,
+      "rewards/accuracy_reward": 0.446875,
+      "rewards/format_reward": 0.6375,
+      "step": 75
+    },
+    {
+      "completion_length": 51.165625,
+      "epoch": 0.012794882047181128,
+      "grad_norm": 4.4375,
+      "kl": 0.12180936294025742,
+      "learning_rate": 2.610135839491878e-06,
+      "loss": 0.0049,
+      "reward": 1.053125,
+      "reward_std": 0.5347494941204787,
+      "rewards/accuracy_reward": 0.4375,
+      "rewards/format_reward": 0.615625,
+      "step": 80
+    },
+    {
+      "completion_length": 47.96875,
+      "epoch": 0.013594562175129948,
+      "grad_norm": 5.28125,
+      "kl": 0.14888262754539028,
+      "learning_rate": 2.538707582879289e-06,
+      "loss": 0.006,
+      "reward": 1.134375,
+      "reward_std": 0.5877825018018484,
+      "rewards/accuracy_reward": 0.4625,
+      "rewards/format_reward": 0.671875,
+      "step": 85
+    },
+    {
+      "completion_length": 47.74375,
+      "epoch": 0.014394242303078768,
+      "grad_norm": 4.84375,
+      "kl": 0.12864484217716382,
+      "learning_rate": 2.4624363920282416e-06,
+      "loss": 0.0051,
+      "reward": 1.0875,
+      "reward_std": 0.6187184229493141,
+      "rewards/accuracy_reward": 0.440625,
+      "rewards/format_reward": 0.646875,
+      "step": 90
+    },
+    {
+      "completion_length": 46.46875,
+      "epoch": 0.015193922431027589,
+      "grad_norm": 4.1875,
+      "kl": 0.12670737880980595,
+      "learning_rate": 2.3816778784387097e-06,
+      "loss": 0.0051,
+      "reward": 1.11875,
+      "reward_std": 0.5038135729730129,
+      "rewards/accuracy_reward": 0.434375,
+      "rewards/format_reward": 0.684375,
+      "step": 95
+    },
+    {
+      "completion_length": 48.203125,
+      "epoch": 0.01599360255897641,
+      "grad_norm": 4.84375,
+      "kl": 0.13195679314667358,
+      "learning_rate": 2.296808575580705e-06,
+      "loss": 0.0053,
+      "reward": 1.121875,
+      "reward_std": 0.5347494941204787,
+      "rewards/accuracy_reward": 0.434375,
+      "rewards/format_reward": 0.6875,
+      "step": 100
+    },
+    {
+      "completion_length": 57.7125,
+      "epoch": 0.01679328268692523,
+      "grad_norm": 4.375,
+      "kl": 0.13236376940039918,
+      "learning_rate": 2.208224183321428e-06,
+      "loss": 0.0053,
+      "reward": 1.1875,
+      "reward_std": 0.49497473835945127,
+      "rewards/accuracy_reward": 0.49375,
+      "rewards/format_reward": 0.69375,
+      "step": 105
+    },
+    {
+      "completion_length": 47.271875,
+      "epoch": 0.01759296281487405,
+      "grad_norm": 5.1875,
+      "kl": 0.1403195610502735,
+      "learning_rate": 2.1163377229898226e-06,
+      "loss": 0.0056,
+      "reward": 1.190625,
+      "reward_std": 0.49939415566623213,
+      "rewards/accuracy_reward": 0.50625,
+      "rewards/format_reward": 0.684375,
+      "step": 110
+    },
+    {
+      "completion_length": 50.625,
+      "epoch": 0.018392642942822873,
+      "grad_norm": 6.34375,
+      "kl": 0.12818894329247996,
+      "learning_rate": 2.0215776116804834e-06,
+      "loss": 0.0051,
+      "reward": 1.175,
+      "reward_std": 0.548007746040821,
+      "rewards/accuracy_reward": 0.46875,
+      "rewards/format_reward": 0.70625,
+      "step": 115
+    },
+    {
+      "completion_length": 49.83125,
+      "epoch": 0.01919232307077169,
+      "grad_norm": 5.28125,
+      "kl": 0.1347155719064176,
+      "learning_rate": 1.924385664775395e-06,
+      "loss": 0.0054,
+      "reward": 1.065625,
+      "reward_std": 0.49055532105267047,
+      "rewards/accuracy_reward": 0.425,
+      "rewards/format_reward": 0.640625,
+      "step": 120
+    },
+    {
+      "completion_length": 51.440625,
+      "epoch": 0.019992003198720514,
+      "grad_norm": 5.34375,
+      "kl": 0.14926065368345007,
+      "learning_rate": 1.8252150359966712e-06,
+      "loss": 0.006,
+      "reward": 1.125,
+      "reward_std": 0.49497473835945127,
+      "rewards/accuracy_reward": 0.46875,
+      "rewards/format_reward": 0.65625,
+      "step": 125
+    },
+    {
+      "epoch": 0.020471811275489804,
+      "eval_completion_length": 49.375,
+      "eval_kl": 0.1358852016114413,
+      "eval_loss": 0.005289154592901468,
+      "eval_reward": 1.1934931506849316,
+      "eval_reward_std": 0.5254868787445434,
+      "eval_rewards/accuracy_reward": 0.464041095890411,
+      "eval_rewards/format_reward": 0.7294520547945206,
+      "eval_runtime": 21.8894,
+      "eval_samples_per_second": 4.568,
+      "eval_steps_per_second": 1.142,
+      "step": 128
+    },
+    {
+      "completion_length": 47.703125,
+      "epoch": 0.020791683326669332,
+      "grad_norm": 5.03125,
+      "kl": 0.1484105174895376,
+      "learning_rate": 1.7245281045947165e-06,
+      "loss": 0.0057,
+      "reward": 1.15625,
+      "reward_std": 0.59662133641541,
+      "rewards/accuracy_reward": 0.515625,
+      "rewards/format_reward": 0.640625,
+      "step": 130
+    },
+    {
+      "completion_length": 48.54375,
+      "epoch": 0.021591363454618154,
+      "grad_norm": 6.5,
+      "kl": 0.1305572761921212,
+      "learning_rate": 1.62279431952272e-06,
+      "loss": 0.0052,
+      "reward": 1.125,
+      "reward_std": 0.5038135729730129,
+      "rewards/accuracy_reward": 0.425,
+      "rewards/format_reward": 0.7,
+      "step": 135
+    },
+    {
+      "completion_length": 50.9,
+      "epoch": 0.022391043582566973,
+      "grad_norm": 5.5,
+      "kl": 0.13983009678777308,
+      "learning_rate": 1.5204880106489266e-06,
+      "loss": 0.0056,
+      "reward": 1.071875,
+      "reward_std": 0.5435883287340403,
+      "rewards/accuracy_reward": 0.41875,
+      "rewards/format_reward": 0.653125,
+      "step": 140
+    },
+    {
+      "completion_length": 48.909375,
+      "epoch": 0.023190723710515795,
+      "grad_norm": 5.375,
+      "kl": 0.1370915047591552,
+      "learning_rate": 1.418086177211835e-06,
+      "loss": 0.0055,
+      "reward": 1.128125,
+      "reward_std": 0.525910659506917,
+      "rewards/accuracy_reward": 0.459375,
+      "rewards/format_reward": 0.66875,
+      "step": 145
+    },
+    {
+      "completion_length": 46.296875,
+      "epoch": 0.023990403838464614,
+      "grad_norm": 6.0,
+      "kl": 0.14994264962151646,
+      "learning_rate": 1.3160662638295526e-06,
+      "loss": 0.006,
+      "reward": 1.18125,
+      "reward_std": 0.48613590374588966,
+      "rewards/accuracy_reward": 0.49375,
+      "rewards/format_reward": 0.6875,
+      "step": 150
+    },
+    {
+      "completion_length": 50.034375,
+      "epoch": 0.024790083966413436,
+      "grad_norm": 4.5625,
+      "kl": 0.13909590368857608,
+      "learning_rate": 1.2149039344325896e-06,
+      "loss": 0.0056,
+      "reward": 1.0875,
+      "reward_std": 0.49497473835945127,
+      "rewards/accuracy_reward": 0.4125,
+      "rewards/format_reward": 0.675,
+      "step": 155
+    },
+    {
+      "completion_length": 52.584375,
+      "epoch": 0.025589764094362255,
+      "grad_norm": 4.46875,
+      "kl": 0.1258360159699805,
+      "learning_rate": 1.11507085449904e-06,
+      "loss": 0.005,
+      "reward": 1.1625,
+      "reward_std": 0.5656854152679444,
+      "rewards/accuracy_reward": 0.46875,
+      "rewards/format_reward": 0.69375,
+      "step": 160
+    },
+    {
+      "completion_length": 50.76875,
+      "epoch": 0.026389444222311077,
+      "grad_norm": 5.875,
+      "kl": 0.17037894912064075,
+      "learning_rate": 1.0170324919323929e-06,
+      "loss": 0.0068,
+      "reward": 1.16875,
+      "reward_std": 0.5038135729730129,
+      "rewards/accuracy_reward": 0.509375,
+      "rewards/format_reward": 0.659375,
+      "step": 165
+    },
+    {
+      "completion_length": 47.490625,
+      "epoch": 0.027189124350259896,
+      "grad_norm": 6.1875,
+      "kl": 0.14714236083673313,
+      "learning_rate": 9.212459468352968e-07,
+      "loss": 0.0059,
+      "reward": 1.103125,
+      "reward_std": 0.5082329902797937,
+      "rewards/accuracy_reward": 0.425,
+      "rewards/format_reward": 0.678125,
+      "step": 170
+    },
+    {
+      "completion_length": 51.25625,
+      "epoch": 0.027988804478208718,
+      "grad_norm": 5.53125,
+      "kl": 0.14223471023142337,
+      "learning_rate": 8.281578202978773e-07,
+      "loss": 0.0057,
+      "reward": 1.175,
+      "reward_std": 0.5126524075865746,
+      "rewards/accuracy_reward": 0.496875,
+      "rewards/format_reward": 0.678125,
+      "step": 175
+    },
+    {
+      "completion_length": 49.021875,
+      "epoch": 0.028788484606157537,
+      "grad_norm": 4.28125,
+      "kl": 0.13608955062227324,
+      "learning_rate": 7.382021321372909e-07,
+      "loss": 0.0054,
+      "reward": 1.215625,
+      "reward_std": 0.4640388172119856,
+      "rewards/accuracy_reward": 0.51875,
+      "rewards/format_reward": 0.696875,
+      "step": 180
+    },
+    {
+      "completion_length": 46.91875,
+      "epoch": 0.02958816473410636,
+      "grad_norm": 4.8125,
+      "kl": 0.13916572104208172,
+      "learning_rate": 6.517982972969913e-07,
+      "loss": 0.0056,
+      "reward": 1.153125,
+      "reward_std": 0.5524271633476019,
+      "rewards/accuracy_reward": 0.4625,
+      "rewards/format_reward": 0.690625,
+      "step": 185
+    },
+    {
+      "completion_length": 47.778125,
+      "epoch": 0.030387844862055178,
+      "grad_norm": 4.375,
+      "kl": 0.14800774758914487,
+      "learning_rate": 5.693491703406479e-07,
+      "loss": 0.0059,
+      "reward": 1.165625,
+      "reward_std": 0.525910659506917,
+      "rewards/accuracy_reward": 0.478125,
+      "rewards/format_reward": 0.6875,
+      "step": 190
+    },
+    {
+      "epoch": 0.030707716913234706,
+      "eval_completion_length": 54.26535087719298,
+      "eval_kl": 0.13607860867979757,
+      "eval_loss": 0.00565306656062603,
+      "eval_reward": 1.1425438596491229,
+      "eval_reward_std": 0.49931662176784714,
+      "eval_rewards/accuracy_reward": 0.49780701754385964,
+      "eval_rewards/format_reward": 0.6447368421052632,
+      "eval_runtime": 21.5975,
+      "eval_samples_per_second": 4.63,
+      "eval_steps_per_second": 1.158,
+      "step": 192
+    },
+    {
+      "completion_length": 59.322916666666664,
+      "epoch": 0.031187524990004,
+      "grad_norm": 6.5625,
+      "kl": 0.14393775458059585,
+      "learning_rate": 4.912391671582093e-07,
+      "loss": 0.0056,
+      "reward": 1.1614583333333333,
+      "reward_std": 0.4345760351667802,
+      "rewards/accuracy_reward": 0.453125,
+      "rewards/format_reward": 0.7083333333333334,
+      "step": 195
+    },
+    {
+      "completion_length": 48.38125,
+      "epoch": 0.03198720511795282,
+      "grad_norm": 5.09375,
+      "kl": 0.15551140770548955,
+      "learning_rate": 4.1783247264156647e-07,
+      "loss": 0.0062,
+      "reward": 1.11875,
+      "reward_std": 0.4684582345187664,
+      "rewards/accuracy_reward": 0.440625,
+      "rewards/format_reward": 0.678125,
+      "step": 200
+    },
+    {
+      "completion_length": 49.403125,
+      "epoch": 0.03278688524590164,
+      "grad_norm": 5.09375,
+      "kl": 0.13503661315189674,
+      "learning_rate": 3.494713426864761e-07,
+      "loss": 0.0054,
+      "reward": 1.296875,
+      "reward_std": 0.5435883287340403,
+      "rewards/accuracy_reward": 0.546875,
+      "rewards/format_reward": 0.75,
+      "step": 205
+    },
+    {
+      "completion_length": 46.8875,
+      "epoch": 0.03358656537385046,
+      "grad_norm": 5.28125,
+      "kl": 0.13574067065492273,
+      "learning_rate": 2.86474508437579e-07,
+      "loss": 0.0054,
+      "reward": 1.2625,
+      "reward_std": 0.5126524075865746,
+      "rewards/accuracy_reward": 0.54375,
+      "rewards/format_reward": 0.71875,
+      "step": 210
+    },
+    {
+      "completion_length": 49.70625,
+      "epoch": 0.03438624550179928,
+      "grad_norm": 4.09375,
+      "kl": 0.12194884136551991,
+      "learning_rate": 2.2913569021667464e-07,
+      "loss": 0.0049,
+      "reward": 1.203125,
+      "reward_std": 0.5435883287340403,
+      "rewards/accuracy_reward": 0.5,
+      "rewards/format_reward": 0.703125,
+      "step": 215
+    },
+    {
+      "completion_length": 49.715625,
+      "epoch": 0.0351859256297481,
+      "grad_norm": 5.28125,
+      "kl": 0.14014763564337046,
+      "learning_rate": 1.7772222806299266e-07,
+      "loss": 0.0056,
+      "reward": 1.203125,
+      "reward_std": 0.4640388172119856,
+      "rewards/accuracy_reward": 0.503125,
+      "rewards/format_reward": 0.7,
+      "step": 220
+    },
+    {
+      "completion_length": 50.23125,
+      "epoch": 0.03598560575769692,
+      "grad_norm": 5.75,
+      "kl": 0.13671348651405424,
+      "learning_rate": 1.3247383527051986e-07,
+      "loss": 0.0055,
+      "reward": 1.209375,
+      "reward_std": 0.48171648643910886,
+      "rewards/accuracy_reward": 0.475,
+      "rewards/format_reward": 0.734375,
+      "step": 225
+    },
+    {
+      "completion_length": 49.9875,
+      "epoch": 0.036785285885645745,
+      "grad_norm": 4.28125,
+      "kl": 0.1529265999328345,
+      "learning_rate": 9.360148073396963e-08,
+      "loss": 0.0061,
+      "reward": 1.13125,
+      "reward_std": 0.5568465806543828,
+      "rewards/accuracy_reward": 0.48125,
+      "rewards/format_reward": 0.65,
+      "step": 230
+    },
+    {
+      "completion_length": 49.721875,
+      "epoch": 0.03758496601359456,
+      "grad_norm": 3.671875,
+      "kl": 0.15454837685683742,
+      "learning_rate": 6.128640531440515e-08,
+      "loss": 0.0062,
+      "reward": 1.109375,
+      "reward_std": 0.49939415566623213,
+      "rewards/accuracy_reward": 0.434375,
+      "rewards/format_reward": 0.675,
+      "step": 235
+    },
+    {
+      "completion_length": 49.690625,
+      "epoch": 0.03838464614154338,
+      "grad_norm": 4.65625,
+      "kl": 0.11877592669334262,
+      "learning_rate": 3.567927681068317e-08,
+      "loss": 0.0048,
+      "reward": 1.2375,
+      "reward_std": 0.5038135729730129,
+      "rewards/accuracy_reward": 0.453125,
+      "rewards/format_reward": 0.784375,
+      "step": 240
+    },
+    {
+      "completion_length": 47.890625,
+      "epoch": 0.039184326269492205,
+      "grad_norm": 5.53125,
+      "kl": 0.17146191628417,
+      "learning_rate": 1.68994874766229e-08,
+      "loss": 0.0069,
+      "reward": 1.15,
+      "reward_std": 0.48613590374588966,
+      "rewards/accuracy_reward": 0.48125,
+      "rewards/format_reward": 0.66875,
+      "step": 245
+    },
+    {
+      "completion_length": 48.56875,
+      "epoch": 0.03998400639744103,
+      "grad_norm": 4.71875,
+      "kl": 0.14072445079218596,
+      "learning_rate": 5.034597359205639e-09,
+      "loss": 0.0056,
+      "reward": 1.2,
+      "reward_std": 0.5391689114272594,
+      "rewards/accuracy_reward": 0.4625,
+      "rewards/format_reward": 0.7375,
+      "step": 250
+    },
+    {
+      "completion_length": 51.275,
+      "epoch": 0.04078368652538984,
+      "grad_norm": 4.84375,
+      "kl": 0.14359070940408855,
+      "learning_rate": 1.3992605321688778e-10,
+      "loss": 0.0057,
+      "reward": 1.209375,
+      "reward_std": 0.4640388172119856,
+      "rewards/accuracy_reward": 0.4875,
+      "rewards/format_reward": 0.721875,
+      "step": 255
+    },
+    {
+      "epoch": 0.04094362255097961,
+      "eval_completion_length": 51.548780487804876,
+      "eval_kl": 0.1350904534593588,
+      "eval_loss": 0.005155267659574747,
+      "eval_reward": 1.1676829268292683,
+      "eval_reward_std": 0.4958370625972748,
+      "eval_rewards/accuracy_reward": 0.47560975609756095,
+      "eval_rewards/format_reward": 0.6920731707317073,
+      "eval_runtime": 24.1797,
+      "eval_samples_per_second": 4.136,
+      "eval_steps_per_second": 1.034,
+      "step": 256
+    },
+    {
+      "epoch": 0.04094362255097961,
+      "step": 256,
+      "total_flos": 0.0,
+      "train_loss": 0.004944845609998083,
+      "train_runtime": 5152.2436,
+      "train_samples_per_second": 3.18,
+      "train_steps_per_second": 0.05
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 256,
+  "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:f4d4914d2c95c3576205435841b4d56e7a276bb9f3bb749a9bdc0d7940323f4c
+size 6264