yitu-opensource / T2T-ViT

ICCV2021, Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Input size is not a square, what should I do in this line?

JerryKingQAQ opened this issue · comments

Thanks to your excellent work!
There has a tensor which size is (3, 56, 112), What should I do in this line to modify ?
self.num_patches = (img_size // (4 * 2 * 2)) * (img_size // (4 * 2 * 2)) # there are 3 sfot split, stride are 4,2,2 seperately