Moritz Wolter
Moritz Wolter
Sure: ``` python from tqdm import tqdm import ptwt, torch from ptwt.wavelets_learnable import ProductFilter torch.manual_seed(42) aten = torch.randn(32, 32, 32, 32) wavelet = ProductFilter(torch.randn(4), torch.randn(4), torch.randn(4), torch.randn(4)) opt = torch.optim.RMSprop(wavelet.parameters(),...
Dear @rachelglenn , You are probably getting a `ValueError: too many values to unpack (expected 2)` because the code assigns a coefficient tuple of length `3` to an expression of...
@rachelglenn, is your question answered?
Closing due to inactivity. Feel free to reopen.
This is a good point. Let's put this on the list.
`float' is difficult to provide at the moment. I ran into error messages from the PyTorch backend for the padding. For complex number support, it's probably a good idea to...
We have some release notes at https://github.com/v0lta/PyTorch-Wavelet-Toolbox/releases. We could include a link to those in the docs.
Having these nice improvements dangling here in the PR makes no sense. I will merge these improvements and open an issue to ensure we don't forget the other improvement ideas.
Okay, it's not that simple, but I think I am getting there.
Work is currently focused on the 1D and 2D cases. When that's done, we can look at 3D, which will take time.