winninghealth/WiNGPT2-Gemma-2-9B-Chat
WiNGPT2-Gemma-2-9B-Chat is a 9 billion parameter, instruction-tuned causal language model developed by winninghealth, based on the Gemma-2 architecture. This model is specifically fine-tuned for the medical domain, excelling in Chinese medical question answering, diagnostic support, and general medical knowledge services. It features a 16384-token context length and demonstrates enhanced performance in Chinese medical evaluations compared to its base model and other general-purpose LLMs.
Loading preview...
Overview
WiNGPT2-Gemma-2-9B-Chat is a 9 billion parameter language model developed by winninghealth, specifically designed for the medical vertical. It is built upon the Gemma-2 architecture and has been instruction-tuned to provide intelligent medical question answering, diagnostic support, and medical knowledge services, aiming to improve diagnostic efficiency and healthcare service quality.
Key Capabilities
- Medical Domain Expertise: Optimized for professional medical knowledge and information integration.
- Enhanced Chinese Medical Performance: Demonstrates a significant improvement (~13%) in Chinese medical capabilities compared to its base model, as measured by the WiNEval-2.0 benchmark.
- Multilingual Support: While primarily focused on Chinese medical enhancement, it also supports multilingual interactions.
- Chat Functionality: Fine-tuned for conversational AI, supporting multi-turn dialogues and translation within the medical context.
- Context Length: Features a substantial context window of 16384 tokens.
Performance Highlights
On the WiNEval 2.0 benchmark, WiNGPT2-Gemma-2-9B-Chat achieved 73.5% on MCKQuiz-2.0 (objective questions) and 82.9% on MSceQA-2.0 (subjective questions). This performance surpasses the base gemma-2-9b-it model (53.7% MCKQuiz, 80.18% MSceQA) and Llama-3.1-8B-Instruct (61.6% MCKQuiz, 73.2% MSceQA) in medical evaluations.
Good for
- Medical Question Answering: Providing accurate answers to medical queries for both general users and professionals.
- Diagnostic Support: Offering informational support for diagnosis and treatment suggestions (for reference only).
- Medical Knowledge Retrieval: Accessing and synthesizing medical information.
- Healthcare AI Applications: Developing intelligent assistants for the healthcare industry.
- Chinese Medical Contexts: Particularly strong in handling Chinese medical language and concepts.