Image Generation on DF-20M
15.12FIDSADM (full)
Evaluation results
| Method | Links | |
|---|---|---|
| SADM (full)Pre-training=ImageNet, Resolution=256x256, Tuning Strategy=Full Fine-tuning2024.02 | 15.12 | |
| SADM (parameter-efficient)Pre-training=ImageNet, Resolution=256x256, Tuning Strategy=Parameter-efficient2024.02 | 16.58 | |
| Full Fine-tuning with DDPMPre-training=ImageNet, Resolution=256x256, Tuning Strategy=Full Fine-tuning2024.02 | 17.26 | |
| DiffFitPre-training=ImageNet, Resolution=256x256, Tuning Strategy=Parameter-efficient2024.02 | 17.35 | |
| BitFitPre-training=ImageNet, Resolution=256x256, Tuning Strategy=Parameter-efficient2024.02 | 17.78 | |
| AdaptFormerPre-training=ImageNet, Resolution=256x256, Tuning Strategy=Parameter-efficient2024.02 | 22.38 | |
| VPTPre-training=ImageNet, Resolution=256x256, Tuning Strategy=Parameter-efficient2024.02 | 32.89 | |
| LoRAPre-training=ImageNet, Resolution=256x256, Tuning Strategy=Parameter-efficient2024.02 | 120.25 |