Multilingual Text Retrieval on xFlickr&CO Full data (2M samples)
54.07Recall@1ToMCLIP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| ToMCLIP2025.10 | 54.07 | 79.98 | 87.67 | |
| ToMCLIP(Ldm)loss=Ldm2025.10 | 54.01 | 80.38 | 88.08 | |
| ToMCLIP(Lta)loss=Lta2025.10 | 53.83 | 79.91 | 87.8 | |
| MCLIP2025.10 | 53.38 | 79.48 | 87.34 | |
| CLIP2025.10 | 16.01 | 28.75 | 35.4 |