Image Classification on ImageNet (test) (Adversarial Robustness L1/L2/Linf Metrics)
68.4Clean AccuracyE-AT
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| E-ATBackbone=XCIT-S-L, Uses extra data for pre-training=false, Source Model=Debenedetti and Troncoso-EPFL, 20222024.02 | 68.4 | 38.1 | 51.8 | 23.8 | 23.4 | |
| RAMPBackbone=XCIT-S-L, Uses extra data for pre-training=false, Source Model=Debenedetti and Troncoso-EPFL, 20222024.02 | 66 | 35.7 | 50.2 | 30 | 29.1 | |
| E-ATBackbone=RN-50-l_inf, Uses extra data for pre-training=false, Source Model=Engstrom et al., 20192024.02 | 58.2 | 26.9 | 39.5 | 18.8 | 17.8 | |
| RAMPBackbone=RN-50-l_inf, Uses extra data for pre-training=false, Source Model=Engstrom et al., 20192024.02 | 55.6 | 25.1 | 38.3 | 22.4 | 20.9 |