Image Classification on CIFAR-100 Split
95.02AccuracyHash
Evaluation results
| Method | Links | |
|---|---|---|
| HashBackbone=ViT-B/16, Pre-trained weights=Sup-21K2026.01 | 95.02 | |
| RanPACBackbone=ViT-B/16, Pre-trained weights=Sup-21K2026.01 | 92.2 | |
| ADAM + AdapterBackbone=ViT-B/16, Pre-trained weights=Sup-21K2026.01 | 87.29 | |
| ADAM + SSFBackbone=ViT-B/16, Pre-trained weights=Sup-21K2026.01 | 85.27 | |
| ADAM + VPT-DBackbone=ViT-B/16, Pre-trained weights=Sup-21K2026.01 | 85.04 | |
| Joint2021.06 | 78.78 | |
| NCL2021.06 | 78.38 | |
| NCL (no opt)Optimization=None2021.06 | 77.88 | |
| OWM2021.06 | 77.07 | |
| KFAC2021.06 | 76.61 | |
| EWC2021.06 | 75.38 | |
| SI2021.06 | 74.84 | |
| TPLRef=ECCV' 20, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=20002021.05 | 65.3 | |
| Layerwise Optimization by Gradient DecompositionProtocol=Class-incremental, Backbone=ResNet34, Replay buffer size=20002021.05 | 65.3 | |
| MUCRef=ECCV' 20, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=20002021.05 | 64.7 | |
| GDRef=ICCV' 19, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=20002021.05 | 62.1 | |
| GEMRef=NIPS' 17, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=2000, Re-implemented=True2021.05 | 61.9 | |
| None2021.06 | 61.43 | |
| LUCRRef=CVPR' 19, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=20002021.05 | 61.2 | |
| A-GEMRef=ICLR' 19, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=2000, Re-implemented=True2021.05 | 60.43 | |
| End2EndRef=ECCV' 18, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=20002021.05 | 60.2 | |
| S-GEMRef=ICLR' 19, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=2000, Re-implemented=True2021.05 | 59.98 | |
| DRRef=ECCV' 18, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=20002021.05 | 59.1 | |
| iCarlRef=CVPR' 17, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=20002021.05 | 58.7 | |
| LwFRef=PAMI' 17, Protocol=Class-incremental, Backbone=ResNet34, Replay buffer size=20002021.05 | 58.4 | |
| Joint2021.06 | 49.78 | |
| Joint2021.06 | 46.85 | |
| SSD+EWCK=1, Neurons per layer=1k2025.12 | 40 | |
| NCL2021.06 | 38.79 | |
| SSDK=1, Neurons per layer=1k2025.12 | 35 | |
| SSDK=10, Neurons per layer=1k2025.12 | 34 | |
| NCL (no opt)Optimization=None2021.06 | 32.81 | |
| OWM2021.06 | 29.23 | |
| SDMLPK=10, Neurons per layer=1k2025.12 | 29 | |
| OWM2021.06 | 28.51 | |
| ODDL-S2022.10 | 27.21 | |
| KFAC2021.06 | 26.57 | |
| NCL2021.06 | 26.36 | |
| ODDL2022.10 | 26.2 | |
| SDMLPK=1, Neurons per layer=1k2025.12 | 25 | |
| SI2021.06 | 22.58 | |
| Dynamic-OCM2022.10 | 21.79 | |
| CoPE2022.10 | 21.62 | |
| ER + GMEDvariant=GMED+2022.10 | 20.93 | |
| CNDPM2022.10 | 20.1 | |
| MIR2022.10 | 20 | |
| EWC2021.06 | 19.97 | |
| ER + GMEDvariant=GMED†2022.10 | 19.6 | |
| reservoir2022.10 | 19.57 | |
| None2021.06 | 18.42 | |
| CoPE-CE2022.10 | 18.33 | |
| NCL (no opt)Optimization=None2021.06 | 16.63 | |
| GSS2022.10 | 13.1 | |
| EWCNeurons per layer=1k2025.12 | 12 | |
| GEM2022.10 | 11.12 | |
| iCARL2022.10 | 10.8 | |
| EWC2021.06 | 7.74 | |
| None2021.06 | 7.71 | |
| KFAC2021.06 | 7.59 | |
| SI2021.06 | 7.02 | |
| finetune2022.10 | 3.53 |