Image Classification on ImageNet 1K (test) (Adversarial Robustness)
74.86Top-1 Accuracy (Clean)Vanilla FT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Vanilla FTBackbone=ViT-B/322026.03 | 74.86 | 0 | |
| GRACEBackbone=ViT-B/322026.03 | 74.21 | 25.44 | |
| TPGMBackbone=ViT-B/322026.03 | 74.13 | 0 | |
| SPDBackbone=ViT-B/322026.03 | 73.68 | 0 | |
| FLYPBackbone=ViT-B/322026.03 | 72.15 | 0 | |
| WiSE-FTBackbone=ViT-B/322026.03 | 70.2 | 0 | |
| CLIPBackbone=ViT-B/322026.03 | 63.35 | 0 | |
| PMG-AFTBackbone=ViT-B/322026.03 | 58.2 | 26.8 | |
| LAATBackbone=ViT-B/322026.03 | 55.46 | 21.54 | |
| TeCoABackbone=ViT-B/322026.03 | 52.54 | 24.4 | |
| FAREBackbone=ViT-B/322026.03 | 47.38 | 13.94 |