Image-to-text retrieval on CREPE productivity
17.9R@1 (replace)IL-CLIP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| IL-CLIPPre-training Dataset=CC12M2024.04 | 17.9 | 23.9 | 14.8 | |
| NegCLIPPre-training Dataset=CC12M2024.04 | 14.9 | 35.8 | 15.2 | |
| IL-CLIPPre-training Dataset=CC3M2024.04 | 14.6 | 22.3 | 5.3 | |
| Codebook-CLIPPre-training Dataset=CC12M2024.04 | 14.4 | 22 | 12.8 | |
| CLIPPre-training Dataset=CC12M2024.04 | 12.3 | 19.5 | 14.6 | |
| NegCLIPPre-training Dataset=CC3M2024.04 | 11.6 | 33.3 | 5.8 | |
| Codebook-CLIPPre-training Dataset=CC3M2024.04 | 10.9 | 19.2 | 3.5 | |
| CLIPPre-training Dataset=CC3M2024.04 | 9.8 | 18.1 | 4 |