|
|
--- |
|
|
license: mit |
|
|
datasets: |
|
|
- hongzhouyu/FineMed-SFT |
|
|
- hongzhouyu/FineMed-DPO |
|
|
language: |
|
|
- en |
|
|
- zh |
|
|
base_model: |
|
|
- meta-llama/Llama-3.1-8B |
|
|
- hongzhouyu/FineMedLM |
|
|
library_name: transformers |
|
|
tags: |
|
|
- medical |
|
|
--- |
|
|
|
|
|
<div align="center"> |
|
|
<h1> |
|
|
FineMedLM-o1 |
|
|
</h1> |
|
|
</div> |
|
|
|
|
|
<div align="center"> |
|
|
<a href="https://github.com/hongzhouyu/FineMed" target="_blank">GitHub</a> | <a href="https://arxiv.org/abs/2501.09213" target="_blank">Paper</a> |
|
|
</div> |
|
|
|
|
|
# <span>Introduction</span> |
|
|
**FineMedLM-o1** is a specialized medical LLM engineered for advanced medical reasoning. It employs a multi-step reasoning process, iteratively reflecting on and refining its thought process before delivering a final response. |
|
|
|
|
|
For more information, visit our GitHub repository. |
|
|
|
|
|
# <span>Usage</span> |
|
|
You can use FineMedLM-o1 in the same way as `Llama-3.1-8B-Instruct`: |
|
|
|
|
|
(⚠️**Note**: Please use the system prompt we provide to achieve better inference results.) |
|
|
```python |
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer |
|
|
|
|
|
main_model_name = "yuhongzhou/FineMedLM" |
|
|
model = AutoModelForCausalLM.from_pretrained(main_model_name, device_map="auto") |
|
|
tokenizer = AutoTokenizer.from_pretrained(main_model_name) |
|
|
|
|
|
prompt = ( |
|
|
"""The following are multiple choice questions (with answers) about health. Think step by step and then finish your answer with "the answer is (X)" where X is the correct letter choice. |
|
|
|
|
|
|
|
|
Question: |
|
|
Polio can be eradicated by which of the following? |
|
|
Options: |
|
|
A. Herbal remedies |
|
|
B. Use of antibiotics |
|
|
C. Regular intake of vitamins |
|
|
D. Administration of tetanus vaccine |
|
|
E. Attention to sewage control and hygiene |
|
|
F. Natural immunity acquired through exposure |
|
|
G. Use of antiviral drugs |
|
|
Answer: Let's think step by step. |
|
|
""" |
|
|
) |
|
|
|
|
|
messages = [ |
|
|
{"role": "system", "content": """You are a helpful professional doctor. You need to generate an answer based on the given problem and thoroughly explore the problem through a systematic and long-term thinking process to provide a final and accurate solution. This requires a comprehensive cycle of analysis, summary, exploration, re-evaluation, reflection, backtracking and iteration to form a thoughtful thinking process. Use the background information provided in the text to assist in formulating the answer. Follow these answer guidelines: |
|
|
1. Please structure your response into two main sections: **Thought** and **Summarization**. |
|
|
2. During the **Thought** phase, think step by step based on the given text content. If the text content is used, it must be expressed. |
|
|
3. During the **Summarization** phase, based on the thinking process in the thinking phase, give the final answer to the question. |
|
|
Here is the question: """}, |
|
|
{"role": "user", "content": prompt} |
|
|
] |
|
|
|
|
|
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True) |
|
|
print(text) |
|
|
|
|
|
model_inputs = tokenizer(text, return_tensors="pt").to(model.device) |
|
|
|
|
|
print("-----start generate-----") |
|
|
generated_ids = model.generate( |
|
|
model_inputs.input_ids, |
|
|
max_new_tokens=2048, |
|
|
eos_token_id=tokenizer.eos_token_id |
|
|
) |
|
|
|
|
|
answer = tokenizer.decode(generated_ids[0], skip_special_tokens=False) |
|
|
print(answer) |
|
|
``` |
|
|
|
|
|
FineMedLM-o1 adopts a *slow-thinking* approach, with outputs formatted as: |
|
|
|
|
|
``` |
|
|
**Thought** |
|
|
[Reasoning process] |
|
|
|
|
|
**Summarization** |
|
|
[Output] |
|
|
``` |
|
|
|
|
|
# <span>Citation</span> |
|
|
``` |
|
|
@misc{yu2025finemedlmo1enhancingmedicalreasoning, |
|
|
title={FineMedLM-o1: Enhancing the Medical Reasoning Ability of LLM from Supervised Fine-Tuning to Test-Time Training}, |
|
|
author={Hongzhou Yu and Tianhao Cheng and Ying Cheng and Rui Feng}, |
|
|
year={2025}, |
|
|
eprint={2501.09213}, |
|
|
archivePrefix={arXiv}, |
|
|
primaryClass={cs.CL}, |
|
|
url={https://arxiv.org/abs/2501.09213}, |
|
|
} |
|
|
``` |
