Image Classification on CIFAR-10 (test) (PGD & Auto Attack Robustness)
88.82Accuracy (Natural)AUGMIX
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| AUGMIXTraining Paradigm=Single-step adversarial training, Checkpoint Type=Final2025.12 | 88.82 | 0 | 0 | |
| BaselineTraining Paradigm=Single-step adversarial training, Checkpoint Type=Final2025.12 | 87.77 | 0 | 0 | |
| RandAugmentTraining Paradigm=Single-step adversarial training, Checkpoint Type=Final2025.12 | 84.63 | 0 | 0 | |
| RandAugment + DOM_DATraining Paradigm=Single-step adversarial training, Checkpoint Type=Final2025.12 | 84.52 | 50.1 | 42.53 | |
| AUGMIX + DOM_DATraining Paradigm=Single-step adversarial training, Checkpoint Type=Final2025.12 | 84.31 | 45.15 | 41.16 | |
| Baseline + DOM_RETraining Paradigm=Single-step adversarial training, Checkpoint Type=Final2025.12 | 71.66 | 47.09 | 17.1 |