Bug related to some objects using/not using GPU

[NirDiamant]

When trying to train from scratch, I encounter the following error:
x = self.scale * x
RuntimeError: Expected object of type torch.FloatTensor but found type torch.cuda.FloatTensor for argument #2 'other'

I believe it has something to do about changing either one of the variable to run on GPU or the other one to run on CPU. yet such change might cause a chain of changes.
I would like to know if you had this problem before, and how should I treat it.
Thanks a lot.

[NirDiamant]

problem solved by setting scale to run on cuda