Continual Learning Accuracy on ImageNet-A
70.8AccuracyJoint
Evaluation results
| Method | Links | |
|---|---|---|
| JointBackbone=ViT-B16, Pre-training=ImageNet21K, Evaluation Protocol=Offline, Initial Classes=2, New Classes per Task=12024.02 | 70.8 | |
| RanDumbBackbone=ViT-B16, Pre-training=ImageNet21K, Evaluation Protocol=Offline, Initial Classes=2, New Classes per Task=12024.02 | 42.2 | |
| RanPAC-impBackbone=ViT-B16, Pre-training=ImageNet21K, Evaluation Protocol=Offline, Initial Classes=2, New Classes per Task=12024.02 | 33.8 |