Abstract Reasoning on MMStar
63.21AccuracyNash
Evaluation Results
| Method | Links | |
|---|---|---|
| NashStrategy=Step-wise verification2026.05 | 63.21 | |
| LLaVA CriticStrategy=Step-wise verification2026.05 | 61.07 | |
| BaseStrategy=Step-wise verification2026.05 | 57.25 | |
| VisionSR1Strategy=Step-wise verification2026.05 | 57.2 | |
| SherlockStrategy=Step-wise verification2026.05 | 54.2 |