/torchcvnn

Pytorch layers for Complex Valued Neural Networks

Primary LanguagePythonMIT LicenseMIT

Complex-Valued Neural Networks (CVNN) - Pytorch

docs pytest PyPI version

This is a library that uses pytorch as a back-end for complex valued neural networks.

It was initially developed by Victor Dhédin and Jérémie Levi during their third year project at CentraleSupélec.

Installation

To install the library, it is simple as :

python -m pip install torchcvnn

Installation for developping

To install when developping the library, within a virtual envrionment, you can :

git clone git@github.com:jeremyfix/torchcvnn.git
python3 -m venv torchcvnn-venv
source torchcvnn-venv/bin/activate
python -m pip install -e torchcvnn

This will install torchcvnn in developper mode.

Releasing a new version

To trigger the pipeline for a new release, you have to tag a commit and to push it on the main branch

[main] git tag x.x.x
[main] git push --tags

This will trigger the ci-cd.yml pipeline which builds the distribution, release it on github and on pypi.

Any commit that is not explicitely tagged with a version number does not trigger the release ci-cd pipeline.

Other projects

You might also be interested in some other projects:

Tensorflow based :

  • cvnn developed by colleagues from CentraleSupélec

Pytorch based :