Image-Text Retrieval on DCI long-text
63.93Top-1 AccuracyHyFL-CLIP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HyFL-CLIPBackbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 63.93 | — | — | |
| DeBias-CLIPBackbone=ViT-L/14, Venue=-2026.02 | 63.4 | — | — | |
| DeBias-CLIPBackbone=ViT-L/14, Venue=-2026.02 | 62.4 | — | — | |
| HiMo-CLIP†Backbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 62.06 | — | — | |
| HiMo-CLIP†Backbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 61.94 | — | — | |
| HyFL-CLIPBackbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 61.92 | — | — | |
| FineLIPBackbone=ViT-L/14, Venue=CVPR252026.02 | 60.8 | — | — | |
| FineLIPBackbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 60.8 | — | — | |
| FineLIPBackbone=ViT-L/14, Venue=CVPR252026.02 | 60.7 | — | — | |
| FineLIPBackbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 60.7 | — | — | |
| HyFL-CLIPBackbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 59 | — | — | |
| HyFL-CLIPBackbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 58.75 | — | — | |
| HiMo-CLIP*Backbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 58.59 | — | — | |
| SmartCLIPBackbone=ViT-L/14, Venue=CVPR252026.02 | 58.5 | — | — | |
| SmartCLIPBackbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 58.5 | — | — | |
| DeBias-CLIPBackbone=ViT-B/16, Venue=-2026.02 | 57.8 | — | — | |
| SmartCLIPBackbone=ViT-L/14, Venue=CVPR252026.02 | 57.6 | — | — | |
| SmartCLIPBackbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 57.6 | — | — | |
| DeBias-CLIPBackbone=ViT-B/16, Venue=-2026.02 | 57.4 | — | — | |
| FineLIP*Backbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 57.18 | — | — | |
| HiMo-CLIP*Backbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 57 | — | — | |
| TULIPBackbone=ViT-L/14, Venue=ICLR252026.02 | 56.4 | — | — | |
| TULIPBackbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 56.4 | — | — | |
| TULIPBackbone=ViT-L/14, Venue=ICLR252026.02 | 55.7 | — | — | |
| TULIPBackbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 55.7 | — | — | |
| FineLIP*Backbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 55.22 | — | — | |
| Long-CLIP†Backbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 54.25 | — | — | |
| SmartCLIPBackbone=ViT-B/16, Venue=CVPR252026.02 | 53.4 | — | — | |
| SmartCLIPBackbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 53.4 | — | — | |
| SmartCLIPBackbone=ViT-B/16, Venue=CVPR252026.02 | 52.8 | — | — | |
| SmartCLIPBackbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 52.8 | — | — | |
| Long-CLIPBackbone=ViT-L/14, Venue=ECCV242026.02 | 52.7 | — | — | |
| TULIPBackbone=ViT-B/16, Venue=ICLR252026.02 | 50.6 | — | — | |
| TULIPBackbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 50.6 | — | — | |
| TULIPBackbone=ViT-B/16, Venue=ICLR252026.02 | 50.2 | — | — | |
| TULIPBackbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 50.2 | — | — | |
| Long-CLIP†Backbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 48.38 | — | — | |
| Long-CLIPBackbone=ViT-B/16, Venue=ECCV242026.02 | 47 | — | — | |
| Long-CLIP†Backbone=ViT-L-14, Evaluation Protocol=Zero-shot2026.07 | 46.55 | — | — | |
| Long-CLIPBackbone=ViT-L/14, Venue=ECCV242026.02 | 45.5 | — | — | |
| Long-CLIP†Backbone=ViT-B-16, Evaluation Protocol=Zero-shot2026.07 | 42.21 | — | — | |
| Long-CLIPBackbone=ViT-B/16, Venue=ECCV242026.02 | 41.1 | — | — | |
| CLIPBackbone=ViT-L/14, Venue=ICML212026.02 | 36 | — | — | |
| CLIPBackbone=ViT-B/16, Venue=ICML212026.02 | 35.9 | — | — | |
| CLIPBackbone=ViT-L/14, Venue=ICML212026.02 | 33.9 | — | — | |
| CLIPBackbone=ViT-B/16, Venue=ICML212026.02 | 32.7 | — | — | |
| COSMOSData size=3M, Evaluation setting=Zero-shot2026.03 | — | 52.5 | 50.2 | |
| COSMOSData size=30M, Evaluation setting=Zero-shot2026.03 | — | 67.7 | 63.7 | |
| FLAIRData size=3M, Evaluation setting=Zero-shot2026.03 | — | 50.3 | 47.3 | |
| FLAIRData size=30M, Evaluation setting=Zero-shot2026.03 | — | 66.3 | 61.3 | |
| GoldiCLIPData size=3M, Evaluation setting=Zero-shot2026.03 | — | 57.7 | 52.8 | |
| GoldiCLIPData size=30M, Evaluation setting=Zero-shot2026.03 | — | 71.9 | 67.5 | |
| LiTData size=100M, Evaluation setting=Zero-shot2026.03 | — | 40.9 | 41.7 | |
| Long-CLIPData size=400M, Evaluation setting=Zero-shot2026.03 | — | 44.1 | 47.4 | |
| LoTLIPData size=100M, Evaluation setting=Zero-shot2026.03 | — | 61 | 62.1 | |
| OpenCLIPData size=2B, Evaluation setting=Zero-shot2026.03 | — | 55.4 | 56 | |
| PEcoreData size=5B, Evaluation setting=Zero-shot2026.03 | — | 60 | 60.2 | |
| SigLIPData size=10B, Evaluation setting=Zero-shot2026.03 | — | 56.2 | 57.7 | |
| SigLIP 2Data size=10B, Evaluation setting=Zero-shot2026.03 | — | 55.7 | 55.7 |