Reproduce “MedGo: A Chinese Medical Large Language Model” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Haitao Zhang, Bo An
MedGo is a specialized large language model for Chinese medical applications, built on Qwen2-72B and developed by Shanghai East Hospital and the Chinese Academy of Social Sciences. It achieved state-of-the-art performance on the CBLUE 3.0 benchmark and 78.6% accuracy on a clinical QA dataset, and is currently deployed in a hospital for clinical decision support.
Key insights
- Domain-specific pre-training and fine-tuning, coupled with preference alignment, significantly enhance LLM performance for specialized medical tasks in Chinese.
- The integration of both Western medicine and Traditional Chinese Medicine knowledge is crucial for comprehensive medical AI in the Chinese context.
- Real-world clinical deployment and feedback loops are vital for validating and iteratively improving medical AI models.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.