All Tokens Matter: Token Labeling for Training Better Vision Transformers

Open paper
Benchmark Model Rank Results
efficient-vits-on-imagenet-1k-with-lv-vit-sBase (LV-ViT-S)#3Top 1 Accuracy: 83.3GFLOPs: 6.6
image-classification-on-imagenetLV-ViT-L#144Top 1 Accuracy: 86.4%Number of params: 151MGFLOPs: 214.8
image-classification-on-imagenetLV-ViT-M#338Top 1 Accuracy: 84.1%Number of params: 56MGFLOPs: 16
image-classification-on-imagenetLV-ViT-S#428Top 1 Accuracy: 83.3%Number of params: 26MGFLOPs: 6.6
semantic-segmentation-on-ade20kLV-ViT-L (UperNet, MS)#88Validation mIoU: 51.8Params (M): 209

Rank counts only results with a code link.