Is LabelSmoothingCrossEntropy suitable for binary classification?

fastai
1

Running with it now and seems to be working fine, but was wondering if BCELoss would still be better for 2 category (binary) classification? Working with the NIH malaria dataset.
Thanks!