Multimodal Retrieval on MSCOCO 2 modal
0.47GapCLIP (LT)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CLIP (LT)Evaluation protocol=Locked-Tuning (LT), Modality configuration=2 modal2026.01 | 0.47 | 0.34 | 72.5 | |
| NotAGapModality configuration=2 modal2026.01 | 0.17 | 0.11 | 75.6 | |
| CLIP (FT)Evaluation protocol=Fine-Tuning (FT), Modality configuration=2 modal2026.01 | 0.12 | 0.63 | 73.8 | |
| OursModality configuration=2 modal2026.01 | 0.03 | 0.77 | 77.3 |