Hi,
I am trying to implement a customized instance normalization function (with masks). I wonder if torch.nn.functional.instance_norm back-propogates through mean and variance, or I should detach them in forward pass? Thank you very much!
Hi,
I am trying to implement a customized instance normalization function (with masks). I wonder if torch.nn.functional.instance_norm back-propogates through mean and variance, or I should detach them in forward pass? Thank you very much!