X-D-Lab/Sunsimiao

Based on the README, here's a technical summary: Built on Qwen2-7B via LoRA fine-tuning with a curated medical dataset combining clinical literature, diagnostic records, consultation dialogues, and medical knowledge QA—synthesized from open sources and GPT-4 generation, then human-validated and automated-processed. Achieves 30B-class performance on CMB-Exam benchmarks and exceeds 79% accuracy on Chinese national physician, pharmacist, and nurse licensing exams. Available across ModelScope, Hugging Face, and WiseModel with both LoRA adapters and merged weights; integrates with the LLaMA-Efficient-Tuning framework for deployment and fine-tuning.

471 stars. No commits in the last 6 months.