Clinical Question Answering on RVA
81.28AccuracyDRL
Evaluation results
| Method | Links | |
|---|---|---|
| DRLBackbone=Qwen3-8B, top-k=best-performing, Physician Rationales=true2026.02 | 81.28 | |
| DRLBackbone=LLaMA-3.1-8B-Instruct, top-k=best-performing, Physician Rationales=true2026.02 | 65.23 | |
| Qwen3-8B2026.02 | 56.97 | |
| MedReason-8B2026.02 | 53.49 | |
| LLaMA-3.1-8B-Instruct2026.02 | 49.91 | |
| HuatuoGPT-o1-8B2026.02 | 49.54 | |
| MedPRM-8B2026.02 | 47.98 |