Conditional Image Generation on CIFAR100 (20 classes, 100 samples per class)
40.04FIDcGANTransfer
Evaluation results
| Method | Links | |
|---|---|---|
| cGANTransferBackbone=BigGAN, Pre-training dataset=CIFAR100 (80 classes)2021.02 | 40.04 | |
| BSABackbone=BigGAN, Pre-training dataset=CIFAR100 (80 classes), Iteration=full2021.02 | 40.73 | |
| TransferGANBackbone=BigGAN, Pre-training dataset=CIFAR100 (80 classes)2021.02 | 48.67 | |
| BSABackbone=BigGAN, Pre-training dataset=CIFAR100 (80 classes), Iteration=early2021.02 | 50.88 | |
| ScratchBackbone=BigGAN, Pre-training dataset=None2021.02 | 81.9 |