Image-Text Retrieval on Flickr30k (val)
83Text-to-Image Recall@1GoldiCLIP
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GoldiCLIPData size=30M, Evaluation setting=Zero-shot2026.03 | 83 | 95.9 | 94.8 | 99.1 | |
| PEcoreData size=5B, Evaluation setting=Zero-shot2026.03 | 81.3 | 95.4 | 93.5 | 99 | |
| FLAIRData size=30M, Evaluation setting=Zero-shot2026.03 | 81.1 | 94.9 | 94.7 | 99.3 | |
| COSMOSData size=30M, Evaluation setting=Zero-shot2026.03 | 80.3 | 95.3 | 92.9 | 99.4 | |
| SigLIP 2Data size=10B, Evaluation setting=Zero-shot2026.03 | 80 | 94.5 | 91.8 | 98.8 | |
| SigLIPData size=30M, Evaluation setting=Zero-shot2026.03 | 76.7 | 93.2 | 90.5 | 98.3 | |
| SigLIPData size=10B, Evaluation setting=Zero-shot2026.03 | 75.6 | 92.8 | 89.1 | 98.6 | |
| GoldiCLIPData size=3M, Evaluation setting=Zero-shot2026.03 | 73.8 | 91.2 | 88.7 | 97.4 | |
| DreamLIPData size=30M, Evaluation setting=Zero-shot2026.03 | 73.3 | 91.8 | 89.9 | 99 | |
| OpenCLIPData size=2B, Evaluation setting=Zero-shot2026.03 | 71.9 | 90.4 | 87.5 | 97.7 | |
| Long-CLIPData size=400M, Evaluation setting=Zero-shot2026.03 | 70.7 | 90.6 | 85.9 | 98.5 | |
| COSMOSData size=3M, Evaluation setting=Zero-shot2026.03 | 68.6 | 88.7 | 84.1 | 96.5 | |
| FLAIRData size=3M, Evaluation setting=Zero-shot2026.03 | 65.7 | 86.8 | 78.7 | 95.2 | |
| LoTLIPData size=100M, Evaluation setting=Zero-shot2026.03 | 65.2 | 88 | 86.9 | 97.8 | |
| LiTData size=100M, Evaluation setting=Zero-shot2026.03 | 60.9 | 84.6 | 80 | 95.9 | |
| DreamLIPData size=3M, Evaluation setting=Zero-shot2026.03 | 53.6 | 78.4 | 69.2 | 91.5 |