Image Classification on ImageNet-1k (Accuracy and dCor^2)
75.3AccuracyDINO
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DINOBackbone=ResNet-50, Evaluation Protocol=Linear evaluation2025.02 | 75.3 | 0.49 | |
| BYOLBackbone=ResNet-50, Evaluation Protocol=Linear evaluation2025.02 | 74.3 | 0.145 | |
| Barlow TwinsBackbone=ResNet-50, Evaluation Protocol=Linear evaluation2025.02 | 73.2 | 0.128 | |
| VICRegBackbone=ResNet-50, Evaluation Protocol=Linear evaluation2025.02 | 73.2 | 0.13 | |
| SimCLRBackbone=ResNet-50, Evaluation Protocol=Linear evaluation2025.02 | 69.3 | — | |
| ADMBackbone=ResNet-50, Evaluation Protocol=Linear evaluation2025.02 | 65.3 | 0.022 | |
| MoCoBackbone=ResNet-50, Evaluation Protocol=Linear evaluation2025.02 | 60.6 | 0.111 |