Text-Image Retrieval on Fashion-Gen
60.6Rank@5Kaleido-BERT
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Kaleido-BERT2021.03 | 60.6 | 33.88 | 68.59 | 319.52 | |
| ToMA-domain2026.04 | 49.8 | — | — | — | |
| SemiCLIP2026.04 | 49.5 | — | — | — | |
| ToMA2026.04 | 49.5 | — | — | — | |
| OSCAR2021.03 | 49.14 | 25.1 | 56.68 | 251.53 | |
| FashionBERT2021.03 | 46.48 | 26.75 | 55.74 | 251.36 | |
| ImageBERT2021.03 | 45.2 | 24.78 | 55.9 | 241.3 | |
| S-CLIP2026.04 | 40.9 | — | — | — | |
| Soft-PL2026.04 | 37.4 | — | — | — | |
| VILBERT2021.03 | 37.23 | 21.12 | 50.11 | 218.13 | |
| VLBERT2021.03 | 36.48 | 22.63 | 48.52 | 212.84 | |
| CLIPvariant=oracle2026.04 | 35.3 | — | — | — | |
| CLIPvariant=fine-tuned2026.04 | 31.9 | — | — | — | |
| Hard-PL2026.04 | 30.2 | — | — | — | |
| CLIPvariant=original2026.04 | 27.7 | — | — | — | |
| PFAN2021.03 | 20.79 | 6.2 | 31.52 | 101.9 | |
| VSE++2021.03 | 16.89 | 4.6 | 28.99 | 94.16 | |
| SCAN2021.03 | 13 | 4.3 | 22.3 | 87.29 | |
| VSE2021.03 | 12.76 | 4.35 | 20.91 | 75.2 | |
| CLIPversion=original2026.04 | — | 12.5 | — | — | |
| CLIPversion=fine-tuned2026.04 | — | 13 | — | — | |
| Hard-PL2026.04 | — | 13.7 | — | — | |
| S-CLIP2026.04 | — | 18.9 | — | — | |
| SemiCLIP2026.04 | — | 24.8 | — | — | |
| Soft-PL2026.04 | — | 15.4 | — | — | |
| ToMA2026.04 | — | 25 | — | — | |
| ToMA-domain2026.04 | — | 25.1 | — | — |