Image-to-Text Retrieval on Continual Multimodal Retrieval Benchmark
53.7Recall@1DAR
Evaluation results
| Method | Links | |
|---|---|---|
| DARBackbone=ViT-L/142026.05 | 53.7 | |
| DARBackbone=ViT-B/162026.05 | 49.3 | |
| Mod-XBackbone=ViT-L/142026.05 | 45.8 | |
| FTBackbone=ViT-L/142026.05 | 45.5 | |
| DARBackbone=ViT-B/322026.05 | 43.7 | |
| C-CLIPBackbone=ViT-L/142026.05 | 42.3 | |
| FTBackbone=ViT-B/162026.05 | 38.7 | |
| Mod-XBackbone=ViT-B/162026.05 | 38.5 | |
| C-CLIPBackbone=ViT-B/162026.05 | 35.9 | |
| FTBackbone=ViT-B/322026.05 | 31.1 | |
| Mod-XBackbone=ViT-B/322026.05 | 30.6 | |
| C-CLIPBackbone=ViT-B/322026.05 | 29.2 |