Rethinking the Design Principles of Robust Vision Transformer

Robust-Vision-Transformer

Note: Since the model is trained on our private platform, this transferred code has not been tested and may have some bugs. If you meet any problems, feel free to open an issue!

This repository contains PyTorch code for Robust Vision Transformers.

For details see our paper “Rethinking the Design Principles of Robust Vision Transformer”

First, clone the repository locally:

git clone https://github.com/vtddggg/Robust-Vision-Transformer.git

Install PyTorch 1.7.0+ and torchvision 0.8.1+ and pytorch-image-models 0.3.2:

conda install -c pytorch pytorch torchvision
pip install timm==0.3.2

In addition, einops and kornia is required for using this implementation:

pip install einops
pip install kornia

We use 4 nodes with 8 gpus to train RVT-Ti, RVT-S and RVT-B:

Training

RVT-Ti:

python -m torch.distributed.launch --nproc_per_node=8 --nnodes=4 main.py --model rvt_tiny --data-path

 
 

 
To finish reading, please visit source site