Multiple-choice question answering on ProtMCQs Level 3
72.5AccuracyProtRLSearch
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ProtRLSearchBackbone=Qwen3-8B2026.03 | 72.5 | 989 | 24.3 | |
| ProtLLMBackbone=Qwen3-8B2026.03 | 59.1 | 546 | 18.2 | |
| Search-R1Backbone=Qwen3-8B2026.03 | 49.2 | 2,935 | 32.4 | |
| BioMedSearchBackbone=Qwen3-8B2026.03 | 48.1 | 3,572 | 69.1 | |
| BioReasonBackbone=Qwen3-8B2026.03 | 47.4 | 514 | 15.9 | |
| BaselineBackbone=Qwen3-8B2026.03 | 26.1 | 493 | 12.4 |