conda create -n con-adv python=3
conda activate con-adv
conda install pytorch torchvision cudatoolkit=11.0 -c pytorch 
pip install git+https://github.com/fra31/auto-attack
pip install advertorch tensorboardX

# Standard cross-entropy
python train.py --mode ce --augment_type base --dataset <DATASET>

# Adversarial training
python train.py --mode <ADV_TRAIN OPTION> --augment_type <AUGMENT> --dataset <DATASET>
# Example: Standard AT under CIFAR-10
python train.py --mode adv_train --augment_type base --dataset cifar10

# Consistency regularization
python train.py --consistency --mode <ADV_TRAIN OPTION> --augment_type <AUGMENT> --dataset <DATASET>
# Example: Consistency regularization based on standard AT under CIFAR-10
python train.py --consistency --mode adv_train --augment_type autoaug_sche --dataset cifar10 

python eval.py --mode test_clean_acc --dataset <DATASET> --load_path <MODEL_PATH>

python eval.py --mode test_adv_acc --distance <DISTANCE> --epsilon <EPSILON> --alpha <ALPHA> --n_iters <NUM_ITER> --dataset <DATASET> --load_path <MODEL_PATH>

python eval.py --mode test_auto_attack --epsilon <EPSILON> --distance <DISTANCE> --dataset <DATASET> --load_path <MODEL_PATH>

python eval.py --mode test_mce --dataset <DATASET> --load_path <MODEL_PATH>

@inproceedings{tack2022consistency,
  title={Consistency Regularization for Adversarial Robustness},
  author={Jihoon Tack and Sihyun Yu and Jongheon Jeong and Minseon Kim and Sung Ju Hwang and Jinwoo Shin},
  booktitle={AAAI Conference on Artificial Intelligence},
  year={2022}
}