malhajar/meditron-7b-chat: Medical Information LLM
This model is a 7 billion parameter instruction-tuned language model, developed by Mohamad Alhajar. It is a fine-tuned version of epfl-llm/meditron-7b, specifically trained using SFT (Supervised Fine-Tuning) on the Alpaca dataset to enhance its conversational capabilities.
Key Capabilities
- Medical Information Retrieval: Designed to answer explicit questions related to medicine, building upon the specialized knowledge of its base model.
- English Language Support: Primarily focused on processing and generating responses in English.
- Instruction Following: Fine-tuned to follow instructions effectively, making it suitable for chat-based interactions.
Performance Benchmarks
Evaluated on the Open LLM Leaderboard, malhajar/meditron-7b-chat demonstrates competitive performance for its size:
- Average Score: 49.59
- AI2 Reasoning Challenge (25-Shot): 50.77
- HellaSwag (10-Shot): 75.37
- MMLU (5-Shot): 40.49
- TruthfulQA (0-shot): 48.56
- Winogrande (5-shot): 73.16
- GSM8k (5-shot): 9.17
Good For
- Developers requiring an LLM for medical question-answering applications.
- Use cases where a specialized, instruction-tuned model with a focus on medical knowledge is beneficial.
- Integration into systems that need to provide informative responses on health-related topics.