Class-Incremental Learning on CIFAR-100 50 tasks
93.13Accuracy (Last Task)Joint Training
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Joint TrainingBackbone=ViT-B/16-IN21K2026.05 | 93.13 | — | |
| E2-LoRABackbone=ViT-B/16-IN21K2026.05 | 90.7 | 93.86 | |
| TUNABackbone=ViT-B/16-IN21K2026.05 | 89.41 | 93 | |
| LORA-DRSBackbone=ViT-B/16-IN21K2026.05 | 87.29 | 91.29 | |
| MOSBackbone=ViT-B/16-IN21K2026.05 | 85.51 | 91 | |
| EASEBackbone=ViT-B/16-IN21K2026.05 | 82.1 | 87.65 | |
| DualPromptBackbone=ViT-B/16-IN21K2026.05 | 76.66 | 85.18 | |
| L2PBackbone=ViT-B/16-IN21K2026.05 | 73.91 | 81.9 | |
| InfLoRABackbone=ViT-B/16-IN21K2026.05 | 56.65 | 70.29 | |
| CODA-PromptBackbone=ViT-B/16-IN21K2026.05 | 55.45 | 68.39 |