About the entire data for training

[Bomingmiao]

Hello, I find your work so valuable. And your code is brief and high efficient. However, I have a question about the data for training. In your example code, there is only n000050 which is a subset of VGG you provided. If I want to train the adversial examples of all set Bs, what changes should I make in the shell file and the attacks py file? Thank you so much