Hi, thanks for sharing your code. I noticed that in the impair step, the noisy_data list always maps the noise to class 0 instead of the classes that need to be forgotten. Based on the paper, it seems like noisy_data.append((batch[i], torch.tensor(class_num))) should be replaced with noisy_data.append((batch[i], torch.tensor(cls))) because class_num isn't getting updated inside the loop. Please let me know if I'm missing something here.
Hi, thanks for sharing your code. I noticed that in the impair step, the noisy_data list always maps the noise to class 0 instead of the classes that need to be forgotten. Based on the paper, it seems like
noisy_data.append((batch[i], torch.tensor(class_num)))should be replaced withnoisy_data.append((batch[i], torch.tensor(cls)))because class_num isn't getting updated inside the loop. Please let me know if I'm missing something here.