• Python
• pytorch
• torchvision
• numpy
• tqdm

git clone https://github.com/Originofamonia/YOPO-You-Only-Propagate-Once.git
cd YOPO-You-Only-Propagate-Once

Recommended: create a pytorch virtual environment

python3 -m venv --system-site-packages ~/torch  # or other names you like

Activate your virtual environment (venv):

source ~/torch/bin/activate  # replace torch with your own venv name

Install prerequisites in your venv:

pip3 install -r requirements.txt --user

Run MI loss:

cd YOPO-You-Only-Propagate-Once/experiments/cifar10/wide34_pgd10
python train_mi.py  # train
python eval.py  # eval

Go to directory experiments/CIFAR10/wide34.yopo-5-3 run python train.py -d <whcih_gpu>

You can change all the hyper-parameters in config.py. And change network in network.py Runing this code for the first time will dowload the dataset in ./experiments/CIFAR10/data/, you can modify the path in dataset.py