Prior-Sensitive Evaluation on ViLP-P pure-question
69.5ScoreViGOS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ViGOSBackbone=Qwen2.5-VL 3B2026.06 | 69.5 | 90 | |
| OPSDBackbone=Qwen2.5-VL 3B2026.06 | 66.83 | 92 | |
| Visionary-R1-3BMethod Category=RL Methods, Backbone Size=3B2026.06 | 65.17 | 88 | |
| ViGOSBackbone=Qwen2.5-VL 7B2026.06 | 61.67 | 91.67 | |
| Vision-R1-7BMethod Category=RL Methods, Backbone Size=7B2026.06 | 57.83 | 90 | |
| OPSDBackbone=Qwen2.5-VL 7B2026.06 | 57 | 91.67 | |
| BaselineBackbone=Qwen2.5-VL 3B2026.06 | 55.67 | 80.67 | |
| BaselineBackbone=Qwen2.5-VL 7B2026.06 | 37 | 58.67 |