Medical Question Answering on MedQA (Decision-Useful Rate)
89.8Decision-Useful RateHCQR
Evaluation Results
| Method | Links | |
|---|---|---|
| HCQRBackbone=Qwen3 30B2026.03 | 89.8 | |
| HCQRBackbone=Qwen3 4B2026.03 | 83.2 | |
| HCQRBackbone=Average2026.03 | 82.1 | |
| HCQRBackbone=Llama3.1 8B2026.03 | 81.8 | |
| HCQRBackbone=Llama3.2 3B2026.03 | 73.4 | |
| RewritingBackbone=Qwen3 30B2026.03 | 65.6 | |
| RewritingBackbone=Qwen3 4B2026.03 | 62.2 | |
| RewritingBackbone=Average2026.03 | 56.1 | |
| HyDEBackbone=Qwen3 30B2026.03 | 54.6 | |
| HyDEBackbone=Llama3.1 8B2026.03 | 54.5 | |
| HyDEBackbone=Average2026.03 | 53.4 | |
| HyDEBackbone=Qwen3 4B2026.03 | 52.8 | |
| HyDEBackbone=Llama3.2 3B2026.03 | 51.6 | |
| RewritingBackbone=Llama3.1 8B2026.03 | 51.2 | |
| RewritingBackbone=Llama3.2 3B2026.03 | 45.2 | |
| Rerank-RAGBackbone=Llama3.2 3B2026.03 | 44.1 | |
| Rerank-RAGBackbone=Llama3.1 8B2026.03 | 44.1 | |
| Rerank-RAGBackbone=Qwen3 4B2026.03 | 44.1 | |
| Rerank-RAGBackbone=Qwen3 30B2026.03 | 44.1 | |
| Rerank-RAGBackbone=Average2026.03 | 44.1 | |
| Simple RAGBackbone=Llama3.2 3B2026.03 | 30 | |
| Simple RAGBackbone=Llama3.1 8B2026.03 | 30 | |
| Simple RAGBackbone=Qwen3 4B2026.03 | 30 | |
| Simple RAGBackbone=Qwen3 30B2026.03 | 30 | |
| Simple RAGBackbone=Average2026.03 | 30 | |
| MAIN-RAGBackbone=Qwen3 30B2026.03 | 26.7 | |
| MAIN-RAGBackbone=Llama3.1 8B2026.03 | 26.4 | |
| MAIN-RAGBackbone=Average2026.03 | 25.4 | |
| MAIN-RAGBackbone=Qwen3 4B2026.03 | 24.7 | |
| MAIN-RAGBackbone=Llama3.2 3B2026.03 | 23.7 |