Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

`torch.*fft` is deprecated

See original GitHub issue

Getting the warnings

tests/scattering1d/test_torch_backend_1d.py::test_fft[cuda-TorchSkcudaBackend1D]
  /home/jenkins/workspace/kymatio_dev/kymatio/scattering1d/backend/torch_backend.py:12: UserWarning: The function torch.fft is deprecated and will be removed in PyTorch 1.8. Use the new torch.fft module functions, instead, by importing torch.fft and calling torch.fft.fft or torch.fft.fftn. (Triggered internally at  /pytorch/aten/src/ATen/native/SpectralOps.cpp:567.)
    _fft = lambda x: torch.fft(x, 1, normalized=False)


tests/scattering1d/test_torch_backend_1d.py::test_fft[cuda-TorchSkcudaBackend1D]
  /home/jenkins/workspace/kymatio_dev/kymatio/scattering1d/backend/torch_backend.py:13: UserWarning: The function torch.ifft is deprecated and will be removed in a future PyTorch release. Use the new torch.fft module functions, instead, by importing torch.fft and calling torch.fft.ifft or torch.fft.ifftn. (Triggered internally at  /pytorch/aten/src/ATen/native/SpectralOps.cpp:578.)
    _ifft = lambda x: torch.ifft(x, 1, normalized=False)


tests/scattering1d/test_torch_backend_1d.py::test_fft[cuda-TorchSkcudaBackend1D]
  /home/jenkins/workspace/kymatio_dev/kymatio/scattering1d/backend/torch_backend.py:14: UserWarning: The function torch.irfft is deprecated and will be removed in a future PyTorch release. Use the new torch.fft module functions, instead, by importing torch.fft and calling torch.fft.ifft or torch.fft.irfft. (Triggered internally at  /pytorch/aten/src/ATen/native/SpectralOps.cpp:602.)
    _irfft = lambda x: torch.irfft(x, 1, normalized=False, onesided=False)[..., None]

Should probably fix this soonish.

Issue Analytics

State:
Created a year ago
Reactions:1
Comments:7

Top GitHub Comments

1reaction

jandencommented, Jun 20, 2022

Hmm ok. So they deprecated in 1.7 but didn’t provide an upgrade path until 1.8. Not cool…

At any rate, these warnings will go away from our CI once #892 merges so I’ll close this.

0reactions

lostanlencommented, Jun 18, 2022

so … it’s not so trivial. In Torch 1.7, there is a torch.fft module but

>>> torch.fft.fft2
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: module 'torch.fft' has no attribute 'fft2'

(they did not only move functions from torch to torch.fft, they also changed the semantics)

If we really insist on removing this DeprecationWarning, i think that our simplest response is just catch it by means of a warning.simplefilter. Otherwise, i’d have to add at least three more conditional branches to Kymatio (for 1D/2D/3D) …