PolarizationPruning
PolarizationPruning copied to clipboard
Some question about the code
https://github.com/polarizationpruning/PolarizationPruning/blob/f19f3fdf982625d442028fab355d02dabe781515/imagenet/main.py#L878
Dear authors,
Thank you for your awesome work! I have a question about the code (line878 of PolarizationPruning/imagenet/main.py)
In my opinion, the updateBN should be "m.weight.data.add_(sparsity * torch.sign(m.weight.grad.data))" I would appreciate it if you can solve my confusion.
Thank you!