snuh/hari-q2.5-thinking
The snuh/hari-q2.5-thinking model is a Korean Medical Large Language Model (LLM) developed by the Healthcare AI Research Institute (HARI) at Seoul National University Hospital (SNUH). This model is fine-tuned on Korean medical question-answering (QA) style data, enabling robust performance in clinical reasoning and medical education. It achieves 89.2% accuracy on the Korean Medical Licensing Examination (KMLE) and 88.36% on the USMLE QA benchmark, making it suitable for clinical decision support and medical self-assessment tools.
Loading preview...
Korean Medical LLM for QA
The snuh/hari-q2.5-thinking model is a specialized Large Language Model (LLM) developed by the Healthcare AI Research Institute (HARI) at Seoul National University Hospital (SNUH). It is specifically fine-tuned on Korean medical question-answering (QA) style data to excel in clinical reasoning and medical education.
Key Capabilities & Performance
- Medical QA Generation: Optimized for generating accurate and reasoned responses to medical questions in Korean.
- Clinical Reasoning: Designed to support clinical decision-making processes through its QA capabilities.
- Educational Tool: Ideal for medical education, self-assessment, and training scenarios.
- Benchmark Performance:
- Achieves 89.2% accuracy on the Korean Medical Licensing Examination (KMLE) QA benchmark (KorMedMCQA 5-shot).
- Scores 88.36% on the USMLE QA benchmark (MedQA-USMLE 0-shot).
- Data Privacy: Trained exclusively on publicly available and de-identified data, ensuring no real patient information or PII is included.
Training & Ethical Considerations
The model was fine-tuned using a curated corpus of Korean medical QA-style data, including clinical guidelines, academic publications, and exam-style questions. This data reflects realistic diagnostic and therapeutic scenarios. HARI emphasizes ethical AI development, ensuring the model is intended for research and educational purposes only and should not be used for clinical decisions or patient care.