Anonymiserad/SweMedMCQ-LFM2.5-1.2B-Answer-EN-SV

Overview

Anonymiserad/SweMedMCQ-LFM2.5-1.2B-Answer-EN-SV is a 1.2 billion parameter model, fine-tuned from the LiquidAI/LFM2.5-1.2B-Instruct base model. It is specifically adapted for answering medical multiple-choice questions in both English and Swedish, utilizing a 32768 token context length. This model was developed as part of a research effort to benchmark and adapt LLMs for English-Swedish medical question answering, using the SweMedMCQ benchmark.

Key Capabilities

• Bilingual Medical QA: Optimized for medical multiple-choice questions in both English and Swedish.
• Answer-Level Supervision: Trained to directly output the correct answer index (e.g., "A", "B") for MCQs, rather than generating full explanations.
• JSON Output: Designed to return answers in a structured JSON format, facilitating integration into automated systems.

Good For

• Research in Medical QA: Ideal for academic and research purposes focused on evaluating and developing medical question-answering systems.
• Benchmarking: Suitable for use with the SweMedMCQ benchmark and similar datasets for performance evaluation.
• Automated Medical MCQ Systems: Can be integrated into applications requiring precise, direct answers to medical multiple-choice questions in English or Swedish.

Limitations

• This is a research model and not intended for clinical use, diagnosis, or treatment decisions.
• Trained exclusively on exam-style MCQs; its accuracy is lower than larger models, and generated explanations (if any) may be incorrect.
• Swedish training and test data are machine-translated, which may introduce errors.