Cross-modal Retrieval on Long-caption Dataset
70.2Retrieval Score (Word Dropout p=0.5)HyFL-CLIP
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| HyFL-CLIPevaluation_mode=Zero-shot, context_length=Long-caption2026.07 | 70.2 | 68.22 | 76.16 | 32.55 | 39.05 | |
| HiMo-CLIPevaluation_mode=Zero-shot, context_length=Long-caption2026.07 | 58.75 | 64.44 | 72.94 | 28.24 | 34.84 | |
| FineLIPevaluation_mode=Zero-shot, context_length=Long-caption2026.07 | 58.7 | 64.18 | 73.41 | 27.2 | 33.41 | |
| Long-CLIPevaluation_mode=Zero-shot, context_length=Long-caption2026.07 | 48.88 | 55.41 | 61.79 | 26.5 | 31.29 |