Adaptive Sparse ViT: Towards Learnable Adaptive Token Pruning by Fully Exploiting Self-Attention

Open paper
Benchmark Model Rank Results
efficient-vits-on-imagenet-1k-with-lv-vit-sAS-LV-S (70%)#7Top 1 Accuracy: 83.1GFLOPs: 4.6
efficient-vits-on-imagenet-1k-with-lv-vit-sAS-LV-S (60%)#13Top 1 Accuracy: 82.6GFLOPs: 3.9
efficient-vits-on-imagenet-1k-with-deit-sAS-DeiT-S (65%)#16Top 1 Accuracy: 79.6GFLOPs: 3.0
efficient-vits-on-imagenet-1k-with-deit-sAS-DeiT-S (50%)#30Top 1 Accuracy: 78.7GFLOPs: 2.3

Rank counts only results with a code link.