Image Classification on ImageNet (Adversarial Robustness Metrics)
85.01Clean AccuracyFARE
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FARETraining Data=ImageNet-1k, Adversarial Training=2-step PGD, Epsilon=1/2552025.12 | 85.01 | 32.11 | 39.81 | |
| UCATTraining Data=ImageNet-1k, Adversarial Training=2-step PGD, Epsilon=1/2552025.12 | 70.48 | 47.51 | 45.59 | |
| TeCoATraining Data=ImageNet-1k, Adversarial Training=2-step PGD, Epsilon=1/2552025.12 | 63.87 | 41.88 | 48.51 | |
| CLIPTraining Data=ImageNet-1k, Adversarial Training=2-step PGD, Epsilon=1/2552025.12 | 59.15 | 1.72 | 0.04 |