Python code for ICLR 2022 spotlight paper EViT: Expediting Vision Transformers via Token Reorganizations
Python code for ICLR 2022 spotlight paper EViT: Expediting Vision Transformers via Token Reorganizations
EViT-DeiT-S Finetune-30 Reproduce
fine tune code problem
The figure
Performance at extreme prune ratio
training with oracle vit
Training
Other model zoo, plz?
cls_attn = attn[:, :, 0, 1:]
Visualize
When to start fusing image patches