Allow NaN in batch fitting

Open Jhsmit opened this issue 4 years ago • 1 comments

Make the torch.mean calls NaN compatible such that regularizers can be computed without introducing zero padding

Feb 25 '21 09:02 Jhsmit

Torch recently added nanmean: https://pytorch.org/docs/stable/generated/torch.nanmean.html

Apr 06 '22 08:04 Jhsmit