Vision-Language Compositional Reasoning on What's-up
44.2Relation ScoreCLIP-FT
Evaluation results
| Method | Links | |
|---|---|---|
| CLIP-FTBackbone=CLIP ViT-B/162026.06 | 44.2 | |
| MACCO-CLIPBackbone=CLIP ViT-L/142026.06 | 42.4 | |
| CLIP-FTBackbone=CLIP ViT-L/142026.06 | 42 | |
| CLIPBackbone=CLIP ViT-B/162026.06 | 41.9 | |
| Random Chance2026.06 | 41.7 | |
| MACCO-CLIPBackbone=CLIP ViT-B/162026.06 | 41.7 | |
| SigLIPBackbone=SigLIP ViT-B/162026.06 | 41.5 | |
| CLIPBackbone=CLIP ViT-L/142026.06 | 41.2 | |
| MACCO-SigLIPBackbone=SigLIP ViT-B/162026.06 | 40.8 | |
| SigLIP-FTBackbone=SigLIP ViT-B/162026.06 | 40.5 |