I know the topic is a duplicate of previous post, but I still don’t get the exact answer.
I’ve implemented nn.CELoss for binary classification and it did work. Then, could I expect an additional benefit or performance improvement when using nn.BCEWithLogitsLoss?
Thanks!