Medical Reasoning on MedQA ID (in-distribution)
80.8AccuracyPRA (Ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| PRA (Ours)Policy=Qwen3-4B-Instruct2026.04 | 80.8 | |
| RAG + SCPolicy=Qwen3-4B-Instruct2026.04 | 76.7 | |
| CoT + SCPolicy=Qwen3-4B-Instruct2026.04 | 74.8 | |
| CoTPolicy=Qwen3-4B-Instruct2026.04 | 72.7 | |
| RAGPolicy=Qwen3-4B-Instruct2026.04 | 72.2 | |
| DirectPolicy=Qwen3-4B-Instruct2026.04 | 61.6 | |
| Direct + SCPolicy=Qwen3-4B-Instruct2026.04 | 61.3 |