Medical Reasoning on Medbullets OOD
63.6AccuracyPRA (Ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| PRA (Ours)Policy=Qwen3-4B-Instruct2026.04 | 63.6 | |
| CoT + SCPolicy=Qwen3-4B-Instruct2026.04 | 58.7 | |
| RAG + SCPolicy=Qwen3-4B-Instruct2026.04 | 58.4 | |
| CoTPolicy=Qwen3-4B-Instruct2026.04 | 56.5 | |
| RAGPolicy=Qwen3-4B-Instruct2026.04 | 55.7 | |
| DirectPolicy=Qwen3-4B-Instruct2026.04 | 48.8 | |
| Direct + SCPolicy=Qwen3-4B-Instruct2026.04 | 48.7 |