medalpaca/medalpaca-7b
medalpaca/medalpaca-7b is a 7 billion parameter large language model, based on the LLaMA architecture, specifically fine-tuned for medical domain tasks. It excels at medical question-answering and dialogue, leveraging a diverse dataset including Anki flashcards, Wikidoc, StackExchange, and ChatDoctor. This model is designed to improve performance in specialized medical contexts, offering a focused tool for healthcare-related natural language processing.
Loading preview...
MedAlpaca-7B: A Medical Domain LLM
MedAlpaca-7B is a 7 billion parameter large language model built upon the LLaMA architecture, specifically fine-tuned to address medical domain tasks. Its primary objective is to enhance performance in medical question-answering and dialogue generation.
Key Capabilities & Training
This model was trained on a comprehensive and diverse dataset tailored for medical applications. Key data sources include:
- ChatDoctor: A large dataset of 200,000 medical question-answer pairs.
- Wikidoc: Generated medical Q&A pairs from relevant paragraphs and headings.
- Anki Flashcards: Automatically generated questions and answers from medical flashcards.
- StackExchange: Top-rated Q&A pairs from Academia, Bioinformatics, Biology, Fitness, and Health categories.
Performance & Limitations
Evaluations on the Open LLM Leaderboard show an average score of 44.98, with specific scores like 54.1 on ARC (25-shot) and 80.42 on HellaSwag (10-shot). MMLU (5-shot) scored 41.47. It's important to note that the model's efficacy is primarily within the medical domain and its knowledge level is geared towards medical students, not board-certified physicians. It has not been tested in real-world clinical applications and should be treated as a research tool only, not a substitute for professional medical advice.