Multiple-choice question answering on BioMedMCQs Level 2
75.8AccuracyProtRLSearch
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ProtRLSearchBackbone=Qwen3-8B2026.03 | 75.8 | 785 | 13.4 | |
| Search-R1Backbone=Qwen3-8B2026.03 | 70.9 | 2,582 | 27.1 | |
| BioMedSearchBackbone=Qwen3-8B2026.03 | 69.7 | 3,348 | 65.1 | |
| ProtLLMBackbone=Qwen3-8B2026.03 | 60.4 | 534 | 14.9 | |
| BioReasonBackbone=Qwen3-8B2026.03 | 56.2 | 496 | 16.7 | |
| BaselineBackbone=Qwen3-8B2026.03 | 38.7 | 335 | 6.8 |