Image-to-Text Retrieval on Flickr (unseen)
96.7R@1EVA-CLIP-18B
Evaluation results
| Method | Links | |
|---|---|---|
| EVA-CLIP-18BEvaluation Protocol=q_i -> c_t2026.06 | 96.7 | |
| EVA-CLIP-8BEvaluation Protocol=q_i -> c_t2026.06 | 95.6 | |
| ELVA-7BEvaluation Protocol=q_i -> c_t2026.06 | 95.2 | |
| UniIR-CLIPEvaluation Protocol=q_i -> c_t2026.06 | 94.2 | |
| LamRA-Ret-7BEvaluation Protocol=q_i -> c_t2026.06 | 92.7 | |
| Long-CLIP-LEvaluation Protocol=q_i -> c_t2026.06 | 89.3 | |
| E5-VEvaluation Protocol=q_i -> c_t2026.06 | 88.2 | |
| CLIP-LEvaluation Protocol=q_i -> c_t2026.06 | 87.2 | |
| MagicLens-LEvaluation Protocol=q_i -> c_t2026.06 | 84.6 |