Text-to-Image Retrieval on Continual Multimodal Retrieval Benchmark
52.8Recall@1DAR
Evaluation results
| Method | Links | |
|---|---|---|
| DARBackbone=ViT-L/142026.05 | 52.8 | |
| DARBackbone=ViT-B/162026.05 | 47.5 | |
| FTBackbone=ViT-L/142026.05 | 44 | |
| Mod-XBackbone=ViT-L/142026.05 | 44 | |
| DARBackbone=ViT-B/322026.05 | 42.3 | |
| C-CLIPBackbone=ViT-L/142026.05 | 41 | |
| Mod-XBackbone=ViT-B/162026.05 | 36.8 | |
| FTBackbone=ViT-B/162026.05 | 36.2 | |
| C-CLIPBackbone=ViT-B/162026.05 | 34.3 | |
| FTBackbone=ViT-B/322026.05 | 33.4 | |
| Mod-XBackbone=ViT-B/322026.05 | 32.9 | |
| C-CLIPBackbone=ViT-B/322026.05 | 32.1 |