vit-pytorch · Issues· 142 open

I hope the author can incorporate the implementations of FastViT and FastViTHD.

How to train？

NaViT is slower than expected

SimpleViT misleading summary

Problems regarding training 3D Vision transformer : model does not converge

where is the train.py? tks

Apply Tanh activation function to ViT - MLP Head

How can I use a 1D Vision Transformer?

Why not always project out from Attention block?

Allow to use classification token in CvT?

issue VIT + Loss function

The Total params: and Params size (MB) of the model printed by summary are different from the bit_base model in timm library. Theoretically, the same settings should be the same. What is the reason?

MS-COCO training from Imagenet pretrained checkpoint

ViT-Dino for Medical images

Weight Initialization