Image Classification on Oxford-IIIT Pet (Top-1 Clean and Adversarial Accuracy)
85.9Top-1 Accuracy (Clean)WISE-FT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| WISE-FTBackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 85.9 | 0 | |
| GRACEBackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 85.2 | 30.1 | |
| TPGMBackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 85.14 | 0 | |
| SPDBackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 84.84 | 0 | |
| Vanilla FTBackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 84.3 | 0 | |
| FLYPBackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 84.21 | 0 | |
| CLIPBackbone=ViT-B/32, Evaluation Protocol=Zero-shot2026.03 | 84 | 0 | |
| FAREBackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 75.44 | 20.4 | |
| LAATBackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 72.2 | 22.4 | |
| PMG-AFTBackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 72.1 | 31.5 | |
| TeCoABackbone=ViT-B/32, Pre-training Dataset=ImageNet, Evaluation Protocol=Zero-shot2026.03 | 67.07 | 29.1 |