Text-to-image retrieval on Flickr30K 84 (test)
80.3Recall@1COSMOS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| COSMOSPre-training Data=Merged-30M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 80.3 | 95.3 | 97.6 | |
| COSMOSPre-training Data=YFCC15M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 79.6 | 94.2 | 96.8 | |
| SigLIPPre-training Data=Merged-30M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 76.7 | 93.2 | 96.1 | |
| COSMOSPre-training Data=CC12M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 76.2 | 92.9 | 96.3 | |
| CLIPPre-training Data=Merged-30M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 75.9 | 93.5 | 96.7 | |
| LiTPre-training Data=2.5B, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 75.1 | 92.8 | 96.2 | |
| SigLIPPre-training Data=YFCC15M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 73.6 | 92.1 | 95.5 | |
| DreamLIPPre-training Data=Merged-30M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 73.3 | 91.9 | 95.7 | |
| CLIPPre-training Data=YFCC15M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 72.4 | 90.7 | 94.9 | |
| OpenCLIPPre-training Data=LAION-2B, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 71.9 | 90.5 | 94.8 | |
| MetaCLIPPre-training Data=2.5B, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 70.5 | 90.7 | 94.6 | |
| DreamLIPPre-training Data=YFCC15M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 70 | 89.2 | 93.4 | |
| COSMOSPre-training Data=CC3M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 68.6 | 88.7 | 93.4 | |
| DreamLIPPre-training Data=CC12M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 68.3 | 89.4 | 93.4 | |
| OpenCLIPPre-training Data=LAION-400M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 68.1 | 88.6 | 93.3 | |
| OpenCLIPPre-training Data=DataComp-1B, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 67.9 | 88 | 93 | |
| CLIPPre-training Data=CC12M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 66.9 | 87.7 | 92.7 | |
| SigLIPPre-training Data=CC12M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 66.7 | 88.1 | 92.8 | |
| LaSF-CLIPPre-training Data=CC12M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 59.9 | 84.2 | 90.9 | |
| DreamLIPPre-training Data=CC3M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 53.5 | 78.3 | 85.5 | |
| SigLIPPre-training Data=CC3M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 53.1 | 78.5 | 86.1 | |
| CLIPPre-training Data=CC3M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 52.1 | 78.4 | 85.5 | |
| LaCLIPPre-training Data=CC12M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 51.6 | 78.8 | 86.2 | |
| MLLM-APre-training Data=CC3M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 49.3 | 74.8 | 83.1 | |
| SLIPPre-training Data=CC12M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 46.6 | 73.3 | 80.9 | |
| SF-CLIPPre-training Data=YFCC15M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 46.2 | 73.2 | 82.7 | |
| MaskCLIPPre-training Data=YFCC15M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 45.6 | 73.4 | 82.1 | |
| SLIPPre-training Data=YFCC15M, Evaluation Protocol=Zero-shot, Vision Encoder Architecture=ViT-B/162024.12 | 41.3 | 68.7 | 78.6 |