optim_low breaks if some parameter in the model has None gradient

[ksreenivasan]

In the step() method here, if p.grad = None, then this line will break causing the optimizer to crash. However, in several Deep Learning applications, it is common to have some parameters within a layer or even whole layers to be frozen. QPyTorch's optimizer would not be applicable to these cases.

PyTorch's default optimizer has a simple and elegant solution which is to just skip these parameters treating None has equivalent to 0 gradient. We have implemented this solution here. I would like to propose this change to QPyTorch as well.

[Tiiiger]

hi @kamikazekartik thank you for your suggestion! if you can help fix it and submit a PR, I am happy to merge it. Otherwise it might take a while before I get to it.

[ksreenivasan]

Sounds good @Tiiiger! It's just a few lines of code, so I'll make the change and create a PR later today.