Medical Reasoning on Lancet (OOD)
67AccuracyPRA (Ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| PRA (Ours)Policy=Qwen3-4B-Instruct2026.04 | 67 | |
| CoT + SCPolicy=Qwen3-4B-Instruct2026.04 | 63.5 | |
| CoTPolicy=Qwen3-4B-Instruct2026.04 | 62.4 | |
| RAGPolicy=Qwen3-4B-Instruct2026.04 | 62.1 | |
| RAG + SCPolicy=Qwen3-4B-Instruct2026.04 | 61 | |
| DirectPolicy=Qwen3-4B-Instruct2026.04 | 60.4 | |
| Direct + SCPolicy=Qwen3-4B-Instruct2026.04 | 60.2 |