Does this work?

[mflodzinski]

Does this code even work? Apart from the fact that "log()" method in Loss class is gone and I had to replace it with "torch.log()", the main loss function almost always returns "nan" values thus the model is not learning at all.

I would be grateful to here back from the author :)

[reddyn12]

@msalhab96 Is the loss function correct? Isn't the backwards run supposed to be custom?