Class-Incremental Learning on TinyImageNet 10-step
71.74Average AccuracyConDU
Evaluation results
| Method | Links | |
|---|---|---|
| ConDUEvaluation Protocol=FT, Task Agnostic Inference=true2025.03 | 71.74 | |
| ZSCLTask Agnostic Inference=true2025.03 | 71.62 | |
| ConDUEvaluation Protocol=LoRA, Task Agnostic Inference=true2025.03 | 68.8 | |
| LwFEvaluation Protocol=VR, Task Agnostic Inference=true2025.03 | 67.05 | |
| iCaRLTask Agnostic Inference=true2025.03 | 65.97 | |
| CLIPTask Agnostic Inference=true2025.03 | 65.59 | |
| LwFTask Agnostic Inference=true2025.03 | 44 | |
| Continual FTEvaluation Protocol=FT, Task Agnostic Inference=true2025.03 | 41.54 |