why it didn't use softmax in computing multichoice loss

[eveliao]

the input are just logits, not normalized by softmax, why can we directly compute the cross entropy loss with it and y?

pytorch-openai-transformer-lm/loss.py

Line 22 in bfd8e09

clf_losses = self.clf_criterion(clf_logits, Y)

[eveliao]

cross entropy includes log_softmax function, sorry...