LLE (ViT-H/14, MAE, Edge Aug)

Reported on 5 benchmarks across 3 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision3 results

Image ClassificationonObjectNet
Top-1 Accuracy· 2022-12-09
60.78
best: 82.7 (CoCa)
A Whac-A-Mole Dilemma: Shortcuts Come in Multiples Where Mitigating One Amplifies Others arXiv:2212.04825
Domain GeneralizationonImageNet-R
Top-1 Error Rate· 2022-12-09
33.1
best: 3.9 (Model soups (BASIC-L))
A Whac-A-Mole Dilemma: Shortcuts Come in Multiples Where Mitigating One Amplifies Others arXiv:2212.04825
Domain GeneralizationonImageNet-Sketch
Top-1 accuracy· 2022-12-09
53.39
best: 77.18 (Model soups (BASIC-L))
A Whac-A-Mole Dilemma: Shortcuts Come in Multiples Where Mitigating One Amplifies Others arXiv:2212.04825

Methodology2 results

Domain AdaptationonImageNet-R
Top-1 Error Rate· 2022-12-09
33.1
best: 3.9 (Model soups (BASIC-L))
A Whac-A-Mole Dilemma: Shortcuts Come in Multiples Where Mitigating One Amplifies Others arXiv:2212.04825
Domain AdaptationonImageNet-Sketch
Top-1 accuracy· 2022-12-09
53.39
best: 77.18 (Model soups (BASIC-L))
A Whac-A-Mole Dilemma: Shortcuts Come in Multiples Where Mitigating One Amplifies Others arXiv:2212.04825