Compositional Evaluation on SugarCrepe
94.2Add ScoreC2LIP
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| C2LIPMethod Category=Models fine-tuned by us2026.03 | 94.2 | 88.3 | 73.1 | — | |
| DAC-LLMMethod Category=Composition-aware models2026.03 | 93.7 | 89.5 | 74.6 | — | |
| CE-CLIPMethod Category=Composition-aware models2026.03 | 92.9 | 87 | 74.9 | — | |
| DAC-SAMMethod Category=Composition-aware models2026.03 | 91.6 | 87 | 73.5 | — | |
| CLICMethod Category=Composition-aware models2026.03 | 89.8 | 85.8 | 72.5 | — | |
| TripletCLIPMethod Category=Composition-aware models2026.03 | 88.3 | 87 | 69.9 | — | |
| SigLIPBackbone=ViT-B/16, Fine-tuning Dataset=CC3M, Method Category=Models fine-tuned by us2026.03 | 87.9 | 85.6 | 69.7 | — | |
| SigLIPBackbone=ViT-B/162026.03 | 86.5 | 84.1 | 65.8 | — | |
| NegCLIPMethod Category=Composition-aware models2026.03 | 85.8 | 85 | 75.3 | — | |
| FineCLIPMethod Category=Fine-grained models2026.03 | 85.4 | 85.3 | 66.8 | — | |
| SLVC-RMethod Category=Composition-aware models2026.03 | 85.4 | 78.9 | 68.8 | — | |
| FG-CLIPMethod Category=Fine-grained models2026.03 | 84.7 | 85.1 | 69.9 | — | |
| CoN-CLIPMethod Category=Composition-aware models2026.03 | 82.4 | 77.4 | 62.4 | — | |
| FLAIR-3mMethod Category=Fine-grained models2026.03 | 80.6 | 81.4 | 70.9 | — | |
| SLVC-RLMethod Category=Composition-aware models2026.03 | 78.5 | 75.9 | 65.5 | — | |
| CLIP (OpenAI)Backbone=ViT-B/322026.03 | 73 | 80 | 62.7 | — | |
| DreamLIP-3mMethod Category=Fine-grained models2026.03 | 72.9 | 77.5 | 64.2 | — | |
| CLIP (OpenAI)Backbone=ViT-B/162026.03 | 72.7 | 80.4 | 62.5 | — | |
| LLIPMethod Category=Fine-grained models2026.03 | 71.4 | 78.9 | 57.8 | — | |
| IL-CLIPMethod Category=Codebook-based models2026.03 | 51.8 | 52.8 | 55.2 | — | |
| Codebook-CLIPMethod Category=Codebook-based models2026.03 | 50.6 | 54.5 | 47.5 | — | |
| BaseBackbone=CLIP, Training data=Pretrained2026.04 | — | — | — | 73 | |
| BaseBackbone=PE, Training data=Pretrained2026.04 | — | — | — | 84.4 | |
| FTBackbone=CLIP, Training data=COCO2026.04 | — | — | — | 81 | |
| FTBackbone=CLIP, Training data=TROHN-Img2026.04 | — | — | — | 85.5 | |
| FTBackbone=PE, Training data=COCO2026.04 | — | — | — | 85.7 | |
| FTBackbone=PE, Training data=TROHN-Img2026.04 | — | — | — | 90.1 | |
| TFGlobalBackbone=CLIP, Training data=COCO2026.04 | — | — | — | 73.2 | |
| TFGlobalBackbone=CLIP, Training data=TROHN-Img2026.04 | — | — | — | 82.2 | |
| TFGlobalBackbone=PE, Training data=COCO2026.04 | — | — | — | 78 | |
| TFGlobalBackbone=PE, Training data=TROHN-Img2026.04 | — | — | — | 86.9 | |
| TFLocalBackbone=CLIP, Training data=COCO2026.04 | — | — | — | 80.9 | |
| TFLocalBackbone=CLIP, Training data=TROHN-Img2026.04 | — | — | — | 86.3 | |
| TFLocalBackbone=PE, Training data=COCO2026.04 | — | — | — | 84.4 | |
| TFLocalBackbone=PE, Training data=TROHN-Img2026.04 | — | — | — | 89.2 |