Co-training $2^L$ Submodels for Visual Recognition

Open paper
Benchmark Model Rank Results
image-classification-on-imagenetViT-H@224 (cosub)#63Top 1 Accuracy: 88.0%
image-classification-on-imagenetViT-L@224 (cosub)#85Top 1 Accuracy: 87.5%
image-classification-on-imagenetSwin-L@224 (cosub)#106Top 1 Accuracy: 87.1%
image-classification-on-imagenetViT-B@224 (cosub)#158Top 1 Accuracy: 86.3%
image-classification-on-imagenetSwin-B@224 (cosub)#163Top 1 Accuracy: 86.2%
image-classification-on-imagenetConvNeXt-B@224 (cosub)#191Top 1 Accuracy: 85.8%
image-classification-on-imagenetPiT-B@224 (cosub)#192Top 1 Accuracy: 85.8%
image-classification-on-imagenetViT-M@224 (cosub)#261Top 1 Accuracy: 85.0%
image-classification-on-imagenetRegnetY16GF@224 (cosub)#332Top 1 Accuracy: 84.2%
image-classification-on-imagenetViT-S@224 (cosub)#456Top 1 Accuracy: 83.1%

Rank counts only results with a code link.