File structure

• pytorch-cifar
  • resnet implementation with VAECNN expriments as well
• pytorch-adversarial-training
  • adversarial training repor
• pytorch-vae
  • vae implementation in pytorch
• argument.py
  • include the input arguments for example learning rate etc
• adversarial_training_main.py
  • my implementation for adversarial training
• fast_gradient_sign_untargeted.py
  • FGSM attack method
• VAECNN_main.py
  • my implementation for VAE CNN

Commonly used commands

For adversarial training

$ python3 adversarial_training_main.py --data_root . --batch_size=64 --learning_rate=0.05

To train the VAECNN model

$ python3 VAECNN_main.py --saved_file_name=vaeFirstLayerChangedklckpt.pth

log

14 Jul 2022

• implement it to accept model name
• haven't executed on HPC yet