Image-to-Text Retrieval on Simpsons (val)
28.3R@1CLIP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CLIPSetup=oracle2026.04 | 28.3 | 52.3 | |
| ToMA2026.04 | 27.2 | 55.7 | |
| SemiCLIP2026.04 | 23.7 | 54.4 | |
| CLIPSetup=fine-tune2026.04 | 19.3 | 48.2 | |
| S-CLIP2023.05 | 18.4 | 43.9 | |
| Soft-PL2026.04 | 18.4 | 40.4 | |
| CLIPTraining Mode=Fine-tuned2023.05 | 16.7 | 41.4 | |
| Hard-PL2026.04 | 16.7 | 42.1 | |
| CLIPTraining Mode=Original2023.05 | 15.8 | 40.8 | |
| Hard-PL2023.05 | 15.4 | 41.7 | |
| Soft-PL2023.05 | 15.4 | 39.9 | |
| S-CLIP2026.04 | 15.4 | 43.9 | |
| CLIPSetup=original2026.04 | 13.2 | 35.5 |