Model save

README.md

@@ -0,0 +1,67 @@
+---
+library_name: transformers
+model_name: MCP-Qwen2.5-1.5B-Open-R1-GRPO
+tags:
+- generated_from_trainer
+- trl
+- grpo
+licence: license
+---
+
+# Model Card for MCP-Qwen2.5-1.5B-Open-R1-GRPO
+
+This model is a fine-tuned version of [None](https://huggingface.co/None).
+It has been trained using [TRL](https://github.com/huggingface/trl).
+
+## Quick start
+
+```python
+from transformers import pipeline
+
+question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
+generator = pipeline("text-generation", model="AdAstraAbyssoque/MCP-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/bowen_liu-hkust/huggingface/runs/uce3tlfn) 
+
+
+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: 0.16.0.dev0
+- Transformers: 4.49.0
+- Pytorch: 2.5.1
+- Datasets: 3.3.2
+- 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}}
+}
+```

all_results.json

@@ -0,0 +1,8 @@
+{
+    "total_flos": 0.0,
+    "train_loss": -0.0018907431845686265,
+    "train_runtime": 2665.9182,
+    "train_samples": 400,
+    "train_samples_per_second": 0.15,
+    "train_steps_per_second": 0.005
+}

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"
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "total_flos": 0.0,
+    "train_loss": -0.0018907431845686265,
+    "train_runtime": 2665.9182,
+    "train_samples": 400,
+    "train_samples_per_second": 0.15,
+    "train_steps_per_second": 0.005
+}

trainer_state.json

@@ -0,0 +1,238 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9655172413793104,
+  "eval_steps": 500,
+  "global_step": 14,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 396.88172149658203,
+      "epoch": 0.06896551724137931,
+      "grad_norm": 0.3486388921737671,
+      "kl": 0.0,
+      "learning_rate": 1e-05,
+      "loss": -0.0087,
+      "reward": 0.337053582072258,
+      "reward_std": 0.7216125279664993,
+      "rewards/<lambda>": 0.3169643059372902,
+      "rewards/format_reward": 0.020089286845177412,
+      "step": 1
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 402.25001525878906,
+      "epoch": 0.13793103448275862,
+      "grad_norm": 0.31598225235939026,
+      "kl": 0.0,
+      "learning_rate": 2e-05,
+      "loss": -0.0083,
+      "reward": 0.3638393059372902,
+      "reward_std": 0.7311068773269653,
+      "rewards/<lambda>": 0.3526785895228386,
+      "rewards/format_reward": 0.011160714784637094,
+      "step": 2
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 401.2143020629883,
+      "epoch": 0.20689655172413793,
+      "grad_norm": 0.3833157420158386,
+      "kl": 0.008453369140625,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.0091,
+      "reward": 0.3660714477300644,
+      "reward_std": 0.700179249048233,
+      "rewards/<lambda>": 0.3392857238650322,
+      "rewards/format_reward": 0.026785716181620955,
+      "step": 3
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 300.3839416503906,
+      "epoch": 0.27586206896551724,
+      "grad_norm": 131.17250061035156,
+      "kl": 1.0369873046875,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.0823,
+      "reward": 0.486607164144516,
+      "reward_std": 0.8625286668539047,
+      "rewards/<lambda>": 0.4508928880095482,
+      "rewards/format_reward": 0.035714288242161274,
+      "step": 4
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 373.6428680419922,
+      "epoch": 0.3448275862068966,
+      "grad_norm": 3.0509426593780518,
+      "kl": 0.2314453125,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": -0.0147,
+      "reward": 0.6294643133878708,
+      "reward_std": 0.9069008827209473,
+      "rewards/<lambda>": 0.5178571566939354,
+      "rewards/format_reward": 0.11160714738070965,
+      "step": 5
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 426.9553756713867,
+      "epoch": 0.41379310344827586,
+      "grad_norm": 4.6992268562316895,
+      "kl": 0.201904296875,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.0026,
+      "reward": 0.558035746216774,
+      "reward_std": 0.906622901558876,
+      "rewards/<lambda>": 0.5178571566939354,
+      "rewards/format_reward": 0.0401785746216774,
+      "step": 6
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 455.0915298461914,
+      "epoch": 0.4827586206896552,
+      "grad_norm": 0.3951745331287384,
+      "kl": 0.177734375,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.001,
+      "reward": 0.4508928805589676,
+      "reward_std": 0.8413252383470535,
+      "rewards/<lambda>": 0.4508928805589676,
+      "rewards/format_reward": 0.0,
+      "step": 7
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 450.3794860839844,
+      "epoch": 0.5517241379310345,
+      "grad_norm": 0.369406521320343,
+      "kl": 0.164306640625,
+      "learning_rate": 1e-05,
+      "loss": 0.0424,
+      "reward": 0.500000037252903,
+      "reward_std": 0.8601988106966019,
+      "rewards/<lambda>": 0.500000037252903,
+      "rewards/format_reward": 0.0,
+      "step": 8
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 423.6205596923828,
+      "epoch": 0.6206896551724138,
+      "grad_norm": 0.39313390851020813,
+      "kl": 0.155029296875,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": -0.0279,
+      "reward": 0.5223214626312256,
+      "reward_std": 0.8841909915208817,
+      "rewards/<lambda>": 0.5223214626312256,
+      "rewards/format_reward": 0.0,
+      "step": 9
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 438.4888610839844,
+      "epoch": 0.6896551724137931,
+      "grad_norm": 0.3800601661205292,
+      "kl": 0.149169921875,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": -0.0136,
+      "reward": 0.549107164144516,
+      "reward_std": 0.8823030292987823,
+      "rewards/<lambda>": 0.549107164144516,
+      "rewards/format_reward": 0.0,
+      "step": 10
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 398.26341247558594,
+      "epoch": 0.7586206896551724,
+      "grad_norm": 0.3832401931285858,
+      "kl": 0.141845703125,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": -0.0314,
+      "reward": 0.5089286118745804,
+      "reward_std": 0.8833121657371521,
+      "rewards/<lambda>": 0.5089286118745804,
+      "rewards/format_reward": 0.0,
+      "step": 11
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 424.11832427978516,
+      "epoch": 0.8275862068965517,
+      "grad_norm": 0.3395453095436096,
+      "kl": 0.134521484375,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": -0.0237,
+      "reward": 0.5758928954601288,
+      "reward_std": 0.9218239635229111,
+      "rewards/<lambda>": 0.5758928954601288,
+      "rewards/format_reward": 0.0,
+      "step": 12
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 397.5379638671875,
+      "epoch": 0.896551724137931,
+      "grad_norm": 0.3803483247756958,
+      "kl": 0.1341552734375,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": -0.0323,
+      "reward": 0.5625000149011612,
+      "reward_std": 0.9007573574781418,
+      "rewards/<lambda>": 0.5625000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 13
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 413.578125,
+      "epoch": 0.9655172413793104,
+      "grad_norm": 0.33767423033714294,
+      "kl": 0.13232421875,
+      "learning_rate": 0.0,
+      "loss": -0.0033,
+      "reward": 0.5558035969734192,
+      "reward_std": 0.8983689099550247,
+      "rewards/<lambda>": 0.5535714477300644,
+      "rewards/format_reward": 0.0022321429569274187,
+      "step": 14
+    },
+    {
+      "epoch": 0.9655172413793104,
+      "step": 14,
+      "total_flos": 0.0,
+      "train_loss": -0.0018907431845686265,
+      "train_runtime": 2665.9182,
+      "train_samples_per_second": 0.15,
+      "train_steps_per_second": 0.005
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 14,
+  "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": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}