Closed jcwang123 closed 2 years ago
We notice that in the manuscript, Y should not backward the gradients. But in the codes, there are no operation to stop the gradients?
_, pred = torch.max(output, dim=1) is equivalent to pred = torch.argmax(output, dim=1), which has no gradient backpropagated.
_, pred = torch.max(output, dim=1)
pred = torch.argmax(output, dim=1)
thanks!
We notice that in the manuscript, Y should not backward the gradients. But in the codes, there are no operation to stop the gradients?