Image Classification on CIFAR10 (test) (Clean Accuracy and ASR)
85.52Clean Accuracyclean data
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| clean dataSampler=DDIM, Backbone=ResNet2026.06 | 85.52 | 11.04 | — | — | |
| one-to-oneSampler=DDIM, Backbone=ResNet2026.06 | 84.7 | 69.71 | -0.82 | 58.67 | |
| clean dataSampler=DDPM, Backbone=ResNet2026.06 | 84.1 | 11.6 | — | — | |
| same-classSampler=DDPM, Backbone=ResNet2026.06 | 83.09 | 96.46 | -1.01 | 84.86 | |
| all-to-oneSampler=DDPM, Backbone=ResNet2026.06 | 82.79 | 94.97 | -1.31 | 83.37 | |
| one-to-oneSampler=DDPM, Backbone=ResNet2026.06 | 82.71 | 90.56 | -1.39 | 78.96 | |
| same-classSampler=DDIM, Backbone=ResNet2026.06 | 82.28 | 49.2 | -3.24 | 38.16 | |
| all-to-oneSampler=DDIM, Backbone=ResNet2026.06 | 80.78 | 69.4 | -4.74 | 58.36 |