Class-incremental learning on ImageNet first 10 classes 1.0 (test)
83.2Accuracy (Task 1)CWR
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CWRBackbone=ResNet-50, Type=Discriminative2025.10 | 83.2 | 61.6 | 72.4 | |
| DC (Diffusion Classifier)Backbone=DiT-S/2, Type=Generative2025.10 | 78.4 | 73.6 | 76 | |
| A-VARC+Backbone=VAR-d8, Type=Generative2025.10 | 72.4 | 82.4 | 77.4 | |
| Discriminative Baseline (None)Backbone=ResNet-50, Type=Discriminative, Rehearsal=None2025.10 | 0 | 82.4 | 41.2 |