RuntimeError: Input type (torch.cuda.FloatTensor) and weight type (torch.FloatTensor) should be the same

[vietvo89]

Hi NetoPedro

I am trying play around with your code. But I got a bug like below. I do now know why input and weight type are not the same. In your code you did transform twice. Why do you need to do that? How can it be fixed?

Thanks

Traceback (most recent call last):
...
File "./GitHub/Universal-Adversarial-Perturbations-Pytorch/adversarial_perturbation.py", line 86, in generate
r2 = (net(cur_img1).max(1)[1])
...
RuntimeError: Input type (torch.cuda.FloatTensor) and weight type (torch.FloatTensor) should be the same