Medical Question Answering on RaR-Medicine (test)
1,395LengthCHORD-μ
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CHORD-μbackbone=Qwen2.5-7B-Instruct, training=CHORD-μ2025.08 | 1,395 | 80.6 | |
| CHORD-ϕbackbone=Qwen2.5-7B-Instruct, training=CHORD-ϕ2025.08 | 1,128 | 81.3 | |
| Pure RLbackbone=Qwen2.5-7B-Instruct, training=pure RL2025.08 | 685 | 76.8 | |
| Expert Datasource=medical-o1-reasoning dataset2025.08 | 550 | — | |
| Original Modelbackbone=Qwen2.5-7B-Instruct2025.08 | 402 | 67.5 |