Text-to-Image Retrieval on SciCap (val)
14.8Recall@1CLIP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CLIPSetup=oracle2026.04 | 14.8 | 25.7 | |
| ToMA2026.04 | 13.7 | 22 | |
| SemiCLIP2026.04 | 13.6 | 21.9 | |
| S-CLIP2026.04 | 12.6 | 21.1 | |
| Hard-PL2026.04 | 12.2 | 20.1 | |
| Soft-PL2026.04 | 12.2 | 20.5 | |
| S-CLIP2023.05 | 11.1 | 18.8 | |
| CLIPSetup=fine-tune2026.04 | 11.1 | 18.1 | |
| CLIPTraining Mode=Fine-tuned2023.05 | 10.2 | 17.2 | |
| Soft-PL2023.05 | 9.7 | 16.8 | |
| Hard-PL2023.05 | 9.2 | 16.2 | |
| CLIPTraining Mode=Original2023.05 | 9 | 14.5 | |
| CLIPSetup=original2026.04 | 8.8 | 14.2 |