Multiple-choice question answering on ProtMCQs Level 2
77.4AccuracyProtRLSearch
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ProtRLSearchBackbone=Qwen3-8B2026.03 | 77.4 | 706 | 20.4 | |
| ProtLLMBackbone=Qwen3-8B2026.03 | 62.8 | 514 | 16.3 | |
| Search-R1Backbone=Qwen3-8B2026.03 | 58.1 | 2,582 | 27.6 | |
| BioMedSearchBackbone=Qwen3-8B2026.03 | 57.3 | 3,129 | 58 | |
| BioReasonBackbone=Qwen3-8B2026.03 | 50.6 | 503 | 13.3 | |
| BaselineBackbone=Qwen3-8B2026.03 | 30.5 | 468 | 9.6 |