Conditional Image Generation on CIFAR-100 10% (train)
15.71FIDVision-aided Adversarial training
Evaluation results
| Method | Links | |
|---|---|---|
| Vision-aided Adversarial trainingBase Architecture=BigGAN, Number of Vision-Aided Discriminators=32021.12 | 15.71 | |
| Vision-aided Adversarial trainingBase Architecture=BigGAN, Number of Vision-Aided Discriminators=22021.12 | 20.05 | |
| Vision-aided Adversarial trainingBase Architecture=BigGAN, Number of Vision-Aided Discriminators=12021.12 | 23.3 | |
| DiffAugment + CRBase Architecture=BigGAN, Data Augmentation=DiffAugment, Consistency Regularization=true2021.12 | 30.53 | |
| DiffAugmentBase Architecture=BigGAN, Data Augmentation=DiffAugment2021.12 | 39.59 |