Weighting the cross-entropy loss function for binary classification

[Changing default loss functions](http://this post) answers the question i believe. Also fastai loss function are indeed torch objects.

When doing my research it was the first result, hope it will help future visitor too