Masking Padding

1

Hi,

Is masking the padding as shown here (under Trick 3) still relevant in PyTorch 0.4.1, or does the ignore_index parameter of nn.CrossEntropyLoss takes care of that for me?

Thanks!