bymavis / cas_iclr2021 Goto Github PK

Thank you so much for sharing the code for reproducibility.

I am getting the error because idx is ! greater 0.

idx = np.where(label.cpu().numpy() == np.array([0]*data.shape[0]))[0]

It will be great if you help me in resolving the issue.

Hi, thank you for your nice work. I used this for the nature dataset, not on the adversarial examples. It worked well. But I'm wondering how about transfering this to regression network. I did some trials, but didn't work well. Do you have some experiences? Looking forward to your relpy. Thank you so much

could you please upload the pre-trained model of CAS on resnet50?

In your paper, you analysis the magnitudes and the activation frequency of channel-wise activation(Figure1 and 2). I would be very grateful if you can share this part of the code.