Image Classification on CIFAR10 (test) (Representation Accuracy %)
79.99RA (W-MSE)ZePAD
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ZePADMAPE-Branch=CLIP, Adversarial fine-tuning=CIFAR10, Batch size=64, Learning rate=0.00001, Hyperparameter λ=202026.02 | 79.99 | 81.24 | 82.35 | 79.34 | 78.66 | 80.32 | |
| BaselineMAPE-Branch=CLIP, Adversarial fine-tuning=CIFAR10, Batch size=64, Learning rate=0.00001, Hyperparameter λ=202026.02 | 10.09 | 47.13 | 10.71 | 45.01 | 36.86 | 29.96 |