fft-conv-pytorch

Implementation of 1D, 2D, and 3D FFT convolutions in PyTorch. Much faster than direct convolutions for large kernel sizes.

MIT License

Downloads

18.6K

Stars

472

Committers

View Code on GitHub

Ecosystems: PyTorch, Python

Bot releases are hidden (Show)

fft-conv-pytorch - 1.2.0 Latest Release

Published by fkodom about 1 year ago

What's Changed

Add padding=same, support half-precision input by @papkov in https://github.com/fkodom/fft-conv-pytorch/pull/18

New Contributors

@papkov made their first contribution in https://github.com/fkodom/fft-conv-pytorch/pull/18

Full Changelog: https://github.com/fkodom/fft-conv-pytorch/compare/1.1.3...1.2.0

fft-conv-pytorch - 1.1.3

Published by fkodom over 2 years ago

What's Changed

adaptively moves offset to the right device so that gpu can be used by @alexhagen in https://github.com/fkodom/fft-conv-pytorch/pull/17

New Contributors

@alexhagen made their first contribution in https://github.com/fkodom/fft-conv-pytorch/pull/17

Full Changelog: https://github.com/fkodom/fft-conv-pytorch/compare/1.1.2...1.1.3

fft-conv-pytorch - 1.1.2

Published by fkodom over 2 years ago

What's Changed

Now requires torch >= 1.8 to support gradients on dilated convolutions.
Add Benchmark Plots by @fkodom in https://github.com/fkodom/fft-conv-pytorch/pull/15
Bug Fix and Torch Compatibility by @fkodom in https://github.com/fkodom/fft-conv-pytorch/pull/16

Full Changelog: https://github.com/fkodom/fft-conv-pytorch/compare/1.1.1...1.1.2

fft-conv-pytorch - 1.1.1

Published by fkodom over 2 years ago

Fixes an issue with incorrect gradients due to dilation parameter.

What's Changed

Perform dilation with Kronecker product by @aretor in https://github.com/fkodom/fft-conv-pytorch/pull/13
Cleanup Unit Tests by @fkodom in https://github.com/fkodom/fft-conv-pytorch/pull/14

Full Changelog: https://github.com/fkodom/fft-conv-pytorch/compare/1.1.0...1.1.1

fft-conv-pytorch - 1.1.0

Published by fkodom almost 3 years ago

What's Changed

Add dilation parameter and tests by @aretor in https://github.com/fkodom/fft-conv-pytorch/pull/12

New Contributors

@aretor made their first contribution in https://github.com/fkodom/fft-conv-pytorch/pull/12

Full Changelog: https://github.com/fkodom/fft-conv-pytorch/compare/1.0.1...1.1.0

fft-conv-pytorch - 1.0.1

Published by fkodom almost 3 years ago

What's Changed

feat: remove einsum for efficiency by @yoyololicon in https://github.com/fkodom/fft-conv-pytorch/pull/11

New Contributors

@yoyololicon made their first contribution in https://github.com/fkodom/fft-conv-pytorch/pull/11

Full Changelog: https://github.com/fkodom/fft-conv-pytorch/compare/1.0.0...1.0.1

fft-conv-pytorch - 1.0.0

Published by fkodom almost 3 years ago

fft-conv-pytorch - 1.0.0rc0

Published by fkodom almost 3 years ago

Package Rankings

Top 8.33% on Pypi.org

Related Projects

fastxtend

Train fastai models faster (and other useful tools)

torchstat

Model analyzer in PyTorch

26 Oct 2018 1,460

partial_fft

FFT fun

ConvLSTM_pytorch

Implementation of Convolutional LSTM in PyTorch.

04 Aug 2017 1,964

convnet-benchmarks

Easy benchmarking of all publicly accessible implementations of convnets

12 Jul 2014 2,676

theano_fftconv

Convolution op for Theano based on CuFFT using scikits.cuda

tensor4

tensor4 - pytorch to C++ convertor using lightweight templated tensor library

Complex-YOLOv4-Pytorch

The PyTorch Implementation based on YOLOv4 of the paper: "Complex-YOLO: Real-time 3D Object Detec...

03 Jul 2020 1,234

pytorch-toolbelt

PyTorch extensions for fast R&D prototyping and Kaggle farming

15 Mar 2019 1,476

pytorch_FFTNet

A pytorch implementation of FFTNet.

b3-basic-batchsize-benchmark

Experiments for the blog post "No, We Don't Have to Choose Batch Sizes As Powers Of 2"

Efficient-PyTorch

My best practice of training large dataset using PyTorch.

29 Mar 2018 1,081

octconv-pytorch