BianCang-Qwen2.5-7B: A Specialized Traditional Chinese Medicine LLM
BianCang-Qwen2.5-7B is a 7.6 billion parameter instruction-tuned model developed by the QLU-NLP team, built upon the Qwen2.5 base architecture. This model is specifically engineered to advance the application of large language models in Traditional Chinese Medicine (TCM), aiming to assist doctors in diagnosis and patients in self-assessment.
Key Capabilities and Features
- TCM Specialization: BianCang is trained using a two-stage method involving domain knowledge injection, followed by knowledge activation and alignment, focusing on TCM-specific tasks.
- Diagnostic Prowess: It achieves state-of-the-art performance in core TCM tasks such as disease diagnosis and syndrome differentiation.
- Medical Examination Performance: The model demonstrates strong results across various medical license examinations, including MLEC-TCM, MLEC-CWM, MLEC-Clinic, MLEC-PublicHealth, and MLEC-Stomatology.
- Instruction-Tuned: The "-Instruct" variant is fine-tuned for conversational and instruction-following capabilities, as shown in its ability to provide detailed TCM diagnoses from patient descriptions.
When to Use This Model
BianCang-Qwen2.5-7B is ideal for applications requiring deep understanding and generation of Traditional Chinese Medicine knowledge. It can be leveraged for:
- TCM Diagnosis Assistance: Aiding practitioners in identifying TCM diseases and syndromes.
- Patient Self-Assessment: Providing informed responses for individuals seeking to understand their health from a TCM perspective.
- TCM Knowledge Q&A: Answering complex queries related to TCM principles, treatments, and prescriptions.
- Educational Tools: Supporting learning and research in Traditional Chinese Medicine.
It is important to note that while powerful, this model is intended for academic research and assistance, and its outputs should not replace professional medical advice or diagnosis.