Class-Incremental Learning on CIFAR100 10 Steps (test)
91.66Average AccuracySimE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SimEBackbone=CLIP ViT-L/14, Pre-trained Dataset=Laion-2B2026.03 | 91.66 | 86.03 | |
| SimE2026.03 | 85.94 | 77.1 | |
| ZSCL2026.03 | 82.15 | 73.65 | |
| iCaRL2026.03 | 79.35 | 70.97 | |
| LwF-VR2026.03 | 78.81 | 70.75 | |
| Continual-CLIP2026.03 | 75.17 | 66.72 | |
| DER2026.03 | 74.64 | 64.35 | |
| CLIP2026.03 | 74.47 | 65.92 | |
| DyTox2026.03 | 67.33 | 51.68 | |
| LwF2026.03 | 65.86 | 48.04 | |
| Fine-tune2026.03 | 65.46 | 53.23 | |
| UCIR2026.03 | 58.66 | 43.39 |