News for hackers. Stuff that matters.
Re:Scaling Machine Learning at Uber with Michelangelo by marmaduke on Wednesday November 07, 2018 @02:25PM (#18401513)
> PyTorch, you still do have to define the backward function and worry about bookkeeping the gradient, clearing gradient values at the appropriate time, and explicitly calling to calculate these things in verbose optimizer invocation code

I’ve definitely never had to do that. Where do you get this from?

There are no comments on this comment yet.