Expert/Academic Reasoning on MMMU-Pro (Pass@5, Avg@5)
64.81Pass@5ViGOS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ViGOSBackbone=Qwen2.5-VL 7B2026.06 | 64.81 | 36.48 | |
| BaselineBackbone=Qwen2.5-VL 7B2026.06 | 63.85 | 37.41 | |
| OPSDBackbone=Qwen2.5-VL 7B2026.06 | 63.37 | 36.91 | |
| OPSDBackbone=Qwen2.5-VL 3B2026.06 | 57.04 | 26.24 | |
| ViGOSBackbone=Qwen2.5-VL 3B2026.06 | 56.44 | 26.16 | |
| BaselineBackbone=Qwen2.5-VL 3B2026.06 | 55 | 22.55 | |
| Visionary-R1-3BMethod Category=RL Method2026.06 | 52.71 | 27.1 | |
| Vision-R1-7BMethod Category=RL Method2026.06 | 47.29 | 31.48 |