Difference output layer doesn't exist in pytorch version.

[jc5201]

Doesn't difference output layer exist in pytorch version? In tensorflow version I could find 'output_type' parameter, but not in pytorch version. Is there any reason for not including that layer?

[f90]

I think I left it out for simplicity sake since it didn't affect performance that much, but it is not too hard to add it back in. So there's nothing wrong with the difference output layer itself, I just didn't implement it here. Hope that helps!