82 lines
2.8 KiB
Markdown
82 lines
2.8 KiB
Markdown
|
---
|
||
|
|
license: apache-2.0
|
||
|
|
library_name: transformers
|
||
|
|
pipeline_tag: text-generation
|
||
|
|
datasets:
|
||
|
|
- Satori-reasoning/Satori_FT_data
|
||
|
|
base_model:
|
||
|
|
- Qwen/Qwen2.5-Math-7B
|
||
|
|
---
|
||
|
|
|
||
|
|
**Satori-7B-SFT** is the SFT model checkpoint for training our RL model [Satori-7B-Round2](https://huggingface.co/Satori-reasoning/Satori-7B-Round2). **Satori-7B-SFT** is only trained with a small-scale format tuning (FT) stage that helps the base LLM to internalize the COAT reasoning format.
|
||
|
|
|
||
|
|
|
||
|
|
# **Usage**
|
||
|
|
```python
|
||
|
|
|
||
|
|
import os
|
||
|
|
from tqdm import tqdm
|
||
|
|
import torch
|
||
|
|
from vllm import LLM, SamplingParams
|
||
|
|
|
||
|
|
def generate(question_list,model_path):
|
||
|
|
llm = LLM(
|
||
|
|
model=model_path,
|
||
|
|
trust_remote_code=True,
|
||
|
|
tensor_parallel_size=1,
|
||
|
|
)
|
||
|
|
sampling_params = SamplingParams(
|
||
|
|
max_tokens=4096,
|
||
|
|
temperature=0.0,
|
||
|
|
n=1,
|
||
|
|
skip_special_tokens=True # hide special tokens such as "<|continue|>", "<|reflect|>", and "<|explore|>"
|
||
|
|
)
|
||
|
|
outputs = llm.generate(question_list, sampling_params, use_tqdm=True)
|
||
|
|
completions = [[output.text for output in output_item.outputs] for output_item in outputs]
|
||
|
|
return completions
|
||
|
|
|
||
|
|
def prepare_prompt(question):
|
||
|
|
prompt = f"<|im_start|>user\nSolve the following math problem efficiently and clearly.\nPlease reason step by step, and put your final answer within \\boxed{{}}.\nProblem: {question}<|im_end|>\n<|im_start|>assistant\n"
|
||
|
|
return prompt
|
||
|
|
|
||
|
|
def run():
|
||
|
|
model_path = "Satori-reasoning/Satori-7B-SFT"
|
||
|
|
all_problems = [
|
||
|
|
"which number is larger? 9.11 or 9.9?",
|
||
|
|
]
|
||
|
|
completions = generate(
|
||
|
|
[prepare_prompt(problem_data) for problem_data in all_problems],
|
||
|
|
model_path
|
||
|
|
)
|
||
|
|
|
||
|
|
for completion in completions:
|
||
|
|
print(completion[0])
|
||
|
|
if __name__ == "__main__":
|
||
|
|
run()
|
||
|
|
|
||
|
|
```
|
||
|
|
|
||
|
|
# **Resources**
|
||
|
|
We provide our training datasets:
|
||
|
|
- [Full format tuning dataset](https://huggingface.co/datasets/Satori-reasoning/Satori_FT_data) with 300K unique questions.
|
||
|
|
- [RL dataset](https://huggingface.co/datasets/Satori-reasoning/Satori_RL_data) with 550K unique questions.
|
||
|
|
|
||
|
|
Please refer to our blog and research paper for more technical details of Satori.
|
||
|
|
- [Blog](https://satori-reasoning.github.io/blog/satori/)
|
||
|
|
- [Paper](https://arxiv.org/pdf/2502.02508)
|
||
|
|
|
||
|
|
For code, see https://github.com/Satori-reasoning/Satori
|
||
|
|
|
||
|
|
# **Citation**
|
||
|
|
If you find our model and data helpful, please cite our paper:
|
||
|
|
```
|
||
|
|
@misc{shen2025satorireinforcementlearningchainofactionthought,
|
||
|
|
title={Satori: Reinforcement Learning with Chain-of-Action-Thought Enhances LLM Reasoning via Autoregressive Search},
|
||
|
|
author={Maohao Shen and Guangtao Zeng and Zhenting Qi and Zhang-Wei Hong and Zhenfang Chen and Wei Lu and Gregory Wornell and Subhro Das and David Cox and Chuang Gan},
|
||
|
|
year={2025},
|
||
|
|
eprint={2502.02508},
|
||
|
|
archivePrefix={arXiv},
|
||
|
|
primaryClass={cs.CL},
|
||
|
|
url={https://arxiv.org/abs/2502.02508},
|
||
|
|
}
|
||
|
|
```
|