Spatial Reasoning on STARE
43.3ScoreSpatialSSRL
Evaluation Results
| Method | Links | |
|---|---|---|
| SpatialSSRLBackbone=Qwen2.5-VL-7B2026.04 | 43.3 | |
| BaselineBackbone=Qwen2.5-VL-7B2026.04 | 41.8 | |
| SpatialEvoBackbone=Qwen2.5-VL-7B2026.04 | 41.3 | |
| SpatialEvoBackbone=Qwen2.5-VL-3B2026.04 | 36.9 | |
| SpaceRBackbone=Qwen2.5-VL-3B2026.04 | 36.8 | |
| SpatialSSRLBackbone=Qwen2.5-VL-3B2026.04 | 36.8 | |
| SpaceRBackbone=Qwen2.5-VL-7B2026.04 | 36.2 | |
| BaselineBackbone=Qwen2.5-VL-3B2026.04 | 33.1 | |
| SpatialLadderBackbone=Qwen2.5-VL-3B2026.04 | 26.4 | |
| ViLaSRBackbone=Qwen2.5-VL-7B2026.04 | 21.4 |