Multiple-choice question answering on ProtMCQs Level 1
86.9AccuracyProtRLSearch
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ProtRLSearchBackbone=Qwen3-8B2026.03 | 86.9 | 632 | 15.9 | |
| ProtLLMBackbone=Qwen3-8B2026.03 | 72.1 | 448 | 10.5 | |
| Search-R1Backbone=Qwen3-8B2026.03 | 65.6 | 2,376 | 19.3 | |
| BioMedSearchBackbone=Qwen3-8B2026.03 | 61.4 | 3,217 | 53.8 | |
| BioReasonBackbone=Qwen3-8B2026.03 | 58.9 | 365 | 12.6 | |
| BaselineBackbone=Qwen3-8B2026.03 | 35.7 | 432 | 8.9 |