Medical Reasoning on GPQA OOD
64.4AccuracyPRA (Ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| PRA (Ours)Policy=Qwen3-4B-Instruct2026.04 | 64.4 | |
| CoTPolicy=Qwen3-4B-Instruct2026.04 | 60.8 | |
| RAGPolicy=Qwen3-4B-Instruct2026.04 | 59.2 | |
| RAG + SCPolicy=Qwen3-4B-Instruct2026.04 | 54.4 | |
| CoT + SCPolicy=Qwen3-4B-Instruct2026.04 | 51.8 | |
| DirectPolicy=Qwen3-4B-Instruct2026.04 | 51.1 | |
| Direct + SCPolicy=Qwen3-4B-Instruct2026.04 | 50.8 |