DropPos: Pre-Training Vision Transformers by Reconstructing Dropped Positions

Devs

DropPos: Pre-Training Vision Transformers by Reconstructing Dropped Positions | Read Paper on Bytez