Scaling Medical Reasoning Verification via Tool-Integrated Reinforcement Learning

Benchmark Model Rank Results
medical-question-answering-on-medxpertqa-textMed-TIV (Qwen2.5-7B guided search) (extra data)–Accuracy: 16.04
multiple-choice-question-answering-mcqa-on-medmcqaMed-TIV (Qwen2.5-7B guided search) (extra data)–Test Set (Acc-%): 0.6470