Comments (6)
Thanks all for your interest and suggestions. We have supported PyTorch > 1.7.1 in this new commit. Please upgrade/reinstall the focal-frequency-loss
package through pip
or use the newest code in this repo.
Please close the issue if you find the problem is resolved. Thank you. :)
from focal-frequency-loss.
Is authors considering making a adaptation for high pytorch version like this :
photosynthesis-team/piq#253
Thanks,I don't familier with the fft, and want to use the great work in high pytorch version.
from focal-frequency-loss.
I was also looking into doing this transition myself, but like the other poster, am not familiar with the fft methods in Pytorch. In 1.8 the torch.fft
methods were remove and torch.fft is now a full new module.
https://github.com/pytorch/pytorch/wiki/The-torch.fft-module-in-PyTorch-1.7
Most importantly: the backing code and tensors changed to be complex tensors, and so even if you replace the call to torch.rfft()
:
with something like
return torch.fft.fft(y, dim=2, norm='ortho')
Then later the calls to .max() fail:
focal-frequency-loss/focal_frequency_loss/focal_frequency_loss.py
Lines 65 to 68 in 27f99aa
from focal-frequency-loss.
@kjerk There is another unofficial implementation maybe you can refer to. https://github.com/HeonKK/Focal-Frequency-Loss/blob/main/util/ffl.py
from focal-frequency-loss.
thx
from focal-frequency-loss.
Thanks all for your interest and suggestions. We have supported PyTorch > 1.7.1 in this new commit. Please upgrade/reinstall the
focal-frequency-loss
package throughpip
or use the newest code in this repo.Please close the issue if you find the problem is resolved. Thank you. :)
That's great news, thank you!
from focal-frequency-loss.
Related Issues (15)
- Can you share the core code of SPADE experiment? thank you
- Deceive D: Adaptive Pseudo Augmentation for GAN Training with Limited Data HOT 1
- the spectra of images HOT 2
- stylegan2 training config
- Tensorflow Implementation of Focal Frequency Loss HOT 4
- About the calculation of distance HOT 5
- Train problem HOT 2
- code about other algorithms
- Training probelm HOT 2
- a question about ffl value HOT 1
- Welcome update to OpenMMLab 2.0
- About use HOT 1
- Why `clamp` to 0-1 range?
- How did you weight the KL divergence term and the frequency loss term in the VAE training?
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from focal-frequency-loss.