DRDN: Decoupled Representation Dynamic Network for From-Scratch ViT Class-Incremental Learning
Open paper
Benchmark
Model
Rank
Results
incremental-learning-on-cifar-100-b0-5steps-of-20-classes
DRDN
–
Average Incremental Accuracy: 77.91
incremental-learning-on-imagenet-10-steps
DRDN
–
Average Incremental Accuracy: 71.83
Final Accuracy: 63.81
…
Rank counts only results with a code link.