Long sequence incremental learning on ImageNet-R Split (test)
0.7657FAAHash
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HashBackbone=ViT-B/16, Pre-training=Sup-21K, Incremental Tasks=202026.01 | 0.7657 | 4.62 | |
| CPromptBackbone=ViT-B/16, Pre-training=Sup-21K, Incremental Tasks=202026.01 | 0.7479 | 7.34 | |
| NoRGaBackbone=ViT-B/16, Pre-training=Sup-21K, Incremental Tasks=202026.01 | 0.7432 | 5.14 | |
| HiDe-PromptBackbone=ViT-B/16, Pre-training=Sup-21K, Incremental Tasks=202026.01 | 0.7398 | 4.98 | |
| CODA-PromptBackbone=ViT-B/16, Pre-training=Sup-21K, Incremental Tasks=202026.01 | 0.6996 | 6.79 | |
| L2PBackbone=ViT-B/16, Pre-training=Sup-21K, Incremental Tasks=202026.01 | 0.6972 | 5.92 | |
| DualPromptBackbone=ViT-B/16, Pre-training=Sup-21K, Incremental Tasks=202026.01 | 0.6672 | 5.63 |