stravoris/medmcq-cell-biology-qwen3-1.7b
The stravoris/medmcq-cell-biology-qwen3-1.7b model is a 1.7 billion parameter Qwen3-based language model fine-tuned by stravoris. It specializes in generating answers and explanations for Cell Biology & Histology medical multiple-choice questions (MCQs). This model is designed as a per-subject answer generator within a larger three-hop MedMCQ pipeline, focusing exclusively on its specific medical domain.
Loading preview...
Overview
This model, stravoris/medmcq-cell-biology-qwen3-1.7b, is a specialized 1.7 billion parameter Qwen3-based language model developed by stravoris. It is fine-tuned to address a very specific task: generating correct answers and brief clinical explanations for Cell Biology & Histology medical multiple-choice questions (MCQs).
Key Capabilities
- Specialized MCQ Answering: Given a Cell Biology & Histology topic, an MCQ stem, and four options, the model identifies the correct option and provides a concise explanation.
- Pipeline Component: It functions as the third stage in the MedMCQ three-hop pipeline, which routes medical MCQs through a subject classifier and a topic classifier before reaching this specialized answer generator.
- Domain-Specific: Exclusively focused on Cell Biology & Histology, making it highly targeted for this medical subfield.
Intended Use and Limitations
This model is a sample for demonstration within the MedMCQ project, which explores small, specialized models for medical reasoning. It is not intended for production-grade medical AI systems or clinical decision-making. Users should be aware that it has not been formally evaluated against board-level benchmarks, and its outputs may contain factual errors or outdated information. It is narrow in scope, brittle outside its domain, and sensitive to prompt format. It was trained on a curated educational dataset, inheriting any biases or gaps present in that data.
Training Details
The model was fine-tuned using the Cell Biology & Histology subset of the Stravoris Medical MCQ dataset, which includes educational MCQs with topic labels, stems, options, correct answers, and explanations. The base model is Qwen/Qwen3-1.7B.