Image-to-Text Retrieval on SciCap (val)
14.8R@1CLIP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CLIPSetup=oracle2026.04 | 14.8 | 25.5 | |
| ToMA2026.04 | 13.2 | 21.7 | |
| SemiCLIP2026.04 | 13.1 | 21.5 | |
| S-CLIP2026.04 | 11.6 | 20.6 | |
| Soft-PL2026.04 | 11.5 | 19.4 | |
| Hard-PL2026.04 | 11.2 | 18.9 | |
| CLIPSetup=fine-tune2026.04 | 11 | 19 | |
| S-CLIP2023.05 | 10.9 | 18.8 | |
| CLIPTraining Mode=Fine-tuned2023.05 | 10.5 | 17.9 | |
| Hard-PL2023.05 | 9.7 | 16.8 | |
| CLIPTraining Mode=Original2023.05 | 9.2 | 14.7 | |
| Soft-PL2023.05 | 9.1 | 15.5 | |
| CLIPSetup=original2026.04 | 8.3 | 13.8 |