Image Classification on Standard CLIP Evaluation Suite (IN-1K, Caltech, CUB, DTD, ESAT, FLO, SUN, UCF) (test)
64.7IN-1K AccuracyProAPO
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ProAPOBackbone=ViT-B/32, Optimization Protocol=Automatic Prompt Optimization (w/ 1-shot)2025.02 | 64.7 | 94.4 | 55.4 | 53.5 | 63 | 74.3 | 66.6 | 69 | 67.6 | |
| iCMBackbone=ViT-B/32, Optimization Protocol=Automatic Prompt Optimization (w/ validation set)2025.02 | 64.5 | 92.7 | 56.1 | 51.4 | 56.3 | 72.2 | 66.2 | 67 | 65.8 | |
| Vanilla CLIPBackbone=ViT-B/32, Optimization Protocol=Zero-shot2025.02 | 62.1 | 91.2 | 51.7 | 42.9 | 43.9 | 66 | 62.1 | 60.9 | 60.1 |