Multiple-choice question answering on BioMedMCQs Level 3
71.7AccuracyProtRLSearch
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ProtRLSearchBackbone=Qwen3-8B2026.03 | 71.7 | 873 | 21.3 | |
| Search-R1Backbone=Qwen3-8B2026.03 | 63.5 | 2,835 | 32.4 | |
| BioMedSearchBackbone=Qwen3-8B2026.03 | 62.4 | 3,987 | 78.9 | |
| ProtLLMBackbone=Qwen3-8B2026.03 | 57.9 | 560 | 16.2 | |
| BioReasonBackbone=Qwen3-8B2026.03 | 49.3 | 528 | 20.5 | |
| BaselineBackbone=Qwen3-8B2026.03 | 29.5 | 379 | 7.6 |