🩺 HuatuoGPT-3-9B

🏠 GitHub | 📄 Paper

Introduction

HuatuoGPT-3-9B is a medical LLM built on Qwen3.5-9B with One-stage Policy Optimization (OnePO). OnePO adapts language models to medicine in a single reinforcement-learning stage, without preceding domain-specific supervised fine-tuning. Teacher responses provide temporary guidance and are retired as the model improves.

We release the training code, medical RL dataset, and 8B rubric grader.

HuatuoGPT-3 requires thinking mode. Keep enable_thinking=True during inference. The model generates reasoning before providing its final answer after </think>.

Model Info

Usage

HuatuoGPT-3-9B can be used like Qwen3.5-9B and deployed with vLLM or SGLang.

For direct text inference, use a Transformers version with Qwen3.5 support (transformers>=5.4.0) and accelerate:

from transformers import AutoModelForImageTextToText, AutoProcessor

model_id = "FreedomIntelligence/HuatuoGPT-3-9B"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    dtype="auto",
    device_map="auto",
).eval()

messages = [{
    "role": "user",
    "content": [{"type": "text", "text": "What are the common causes of chest pain?"}],
}]
inputs = processor.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    enable_thinking=True,
    return_dict=True,
    return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=4096)
response = outputs[0, inputs["input_ids"].shape[-1]:]
print(processor.decode(response, skip_special_tokens=True))

📖 Citation

@inproceedings{chen2026onepo,
  title={OnePO: Direct One-stage Policy Optimization for SFT-free Domain Adaptation},
  author={Chen, Junying and Xie, Xinyuan and Li, Ziniu and Wang, Benyou},
  booktitle={Proceedings of the 43rd International Conference on Machine Learning},
  year={2026}
}