Class-Incremental Learning on ImageNet-R N = 10
80.64AccuracyEASE
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| EASEOptimization Strategy=FO Optimization2025.10 | 80.64 | — | 7.41 | |
| InfLoRAOptimization Strategy=FO Optimization2025.10 | 79.54 | — | 9.01 | |
| Adapter + Cls.Optimization Strategy=FO Optimization2025.10 | 78.89 | — | 7.04 | |
| SD-LoRABackbone=ViT-B/162025.11 | 77.34 | 82.04 | — | |
| PS-LoRABackbone=ViT-B/162025.11 | 77.15 | 82.12 | — | |
| ZO-FCOptimization Strategy=ZO Optimization2025.10 | 75.8 | — | 3.64 | |
| HiDe-PromptBackbone=ViT-B/162025.11 | 74.65 | 78.46 | — | |
| CODA-PromptOptimization Strategy=FO Optimization2025.10 | 74.15 | — | 4.56 | |
| L2POptimization Strategy=FO Optimization2025.10 | 74.14 | — | 6.45 | |
| APER AdapterOptimization Strategy=FO Optimization2025.10 | 73.2 | — | 7.04 | |
| LAE AdapterOptimization Strategy=FO Optimization2025.10 | 71.68 | — | 7.95 | |
| L2PBackbone=ViT-B/162025.11 | 71.26 | 76.13 | — | |
| DualPromptBackbone=ViT-B/162025.11 | 68.22 | 73.81 | — | |
| Learnable Cls.Optimization Strategy=FO Optimization2025.10 | 65.67 | — | 3.67 | |
| SimpleCILOptimization Strategy=Classifier Only2025.10 | 61.99 | — | 8.08 | |
| Full FTBackbone=ViT-B/162025.11 | 60.57 | 72.31 | — |