Multiple-choice question answering on BioMedMCQs Level 1
89.2AccuracyProtRLSearch
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ProtRLSearchBackbone=Qwen3-8B2026.03 | 89.2 | 554 | 10.5 | |
| Search-R1Backbone=Qwen3-8B2026.03 | 76.1 | 2,063 | 19.3 | |
| ProtLLMBackbone=Qwen3-8B2026.03 | 74.8 | 409 | 9.6 | |
| BioMedSearchBackbone=Qwen3-8B2026.03 | 73.3 | 3,029 | 48.1 | |
| BioReasonBackbone=Qwen3-8B2026.03 | 63.9 | 472 | 11.7 | |
| BaselineBackbone=Qwen3-8B2026.03 | 45.8 | 321 | 6.5 |