Classification on CIFAR-10 (train test)
92.34Top-1 AccuracySIGReg
Evaluation Results
| Method | Links | |
|---|---|---|
| SIGRegApproach=optimization, Backbone=ResNet-18, Pretraining Epochs=1000, Evaluation Protocol=Linear-probe, Augmentation Protocol=Matched2026.05 | 92.34 | |
| PEIRAApproach=optimization, Backbone=ResNet-18, Pretraining Epochs=1000, Evaluation Protocol=Linear-probe, Augmentation Protocol=Matched2026.05 | 90.97 | |
| VICRegApproach=optimization, Backbone=ResNet-18, Pretraining Epochs=1000, Evaluation Protocol=Linear-probe, Augmentation Protocol=Matched2026.05 | 90.92 |