yasserrmd/SinaReason-Magistral-2509
SinaReason-Magistral-2509 by yasserrmd is a 24 billion parameter instruction-tuned language model, fine-tuned from mistralai/Magistral-Small-2509. It specializes in step-by-step medical clinical reasoning, generating a transparent chain-of-thought process within tags before providing a clinical summary. This model is designed for educational and professional clinical settings, assisting in understanding and formulating medical logic.
Loading preview...
SinaReason-Magistral-2509: Medical Clinical Reasoning Assistant
SinaReason-Magistral-2509 is a 24 billion parameter instruction-tuned language model developed by yasserrmd, building upon the mistralai/Magistral-Small-2509 base model. Inspired by Ibn Sina (Avicenna), this model is specifically designed to emulate logical, evidence-based reasoning in clinical contexts.
Key Capabilities
- Advanced Clinical Reasoning: Leverages the powerful reasoning capabilities of its base model to analyze complex clinical vignettes.
- Chain-of-Thought (CoT) Output: Uniquely structures its responses to first externalize its detailed reasoning process within
<think>...</think>tags, providing transparency before delivering a conclusion. - Instruction-Tuned for Medical Tasks: Fine-tuned on the
FreedomIntelligence/medical-o1-reasoning-SFTdataset to enhance performance on medical reasoning tasks. - Efficient Fine-Tuning: Developed using the Unsloth library, enabling efficient and memory-optimized fine-tuning of large models on accessible hardware.
- Qualitative Performance: Achieved a qualitative grade of A- across 30 diverse medical reasoning prompts, demonstrating consistent CoT adherence and high clinical accuracy.
Good for
- Medical Professionals: Assisting clinicians in understanding and formulating clinical logic.
- Researchers: Exploring advanced medical reasoning capabilities in AI.
- Medical Students: Supporting educational purposes by providing transparent, step-by-step reasoning for clinical scenarios.
- Developing Clinical Decision Support Tools: As a component for systems requiring explainable medical reasoning (with critical human oversight).
IMPORTANT: This model is a research and educational tool and NOT a substitute for a qualified human medical professional. All outputs must be critically reviewed and independently verified by a human expert.