PyTorch-Ignite PyTorch-Ignite


PyTorch-Ignite is a community-driven open source project developed by a passionate group of contributors. Please read the PyTorch-Ignite Code of Conduct for guidance on how to interact with other community members in a way that makes the community thrive.

GitHub Issue Tracker

GitHub issuse tracker should be used for bug reporting, feature requests, documentation improvements, and user feedbacks.

GitHub Discussions

GitHub Discussions can be used for discussing PyTorch-Ignite related questions and help.


Join us on Discord with our community members to interact with members and maintainers, connect with Ignite users across the globe, and chat about anything.

Join the PyTorch-Ignite Community

To thrive, the PyTorch-Ignite project needs your expertise and enthusiasm. Not a coder? Not a problem! There are many ways to contribute to PyTorch-Ignite.

Please checkout Contribution Guidelines.

About us


This project was started by Alykhan Tejani, Francisco Massa and Joost van Amersfoort in the end of 2017 and was highly inspired by torchnet.

In 2018, the project core team had grown and the first public release was published in June 2018. Since then, several releases have appeared following a ~5 month cycle, with a thriving international community leading the development.


PyTorch-Ignite is a NumFOCUS Affiliated Project and open-source community effort. The Project’s affiliation to PyTorch organization is due in large part to historical reasons and it has the same implication to the project’s governance as the other community projects from the PyTorch Ecosystem.

The decision making process and governance structure of the project is described in the governance document.


The following people are currently core contributors to PyTorch-Ignite’s development and maintenance:

Emeritus Core Developers

The following people have been active contributors in the past, but are no longer active in the project:

Join Core Team

We are looking for motivated contributors to become collaborators and help out with the project. We can start considering a candidate after several successfully merged GitHub pull requests. If you are interested, for more details, please, contact Victor (vfdev-5#4458) on PyTorch-Ignite Discord.

Citing PyTorch-Ignite

If you use PyTorch-Ignite in a scientific publication, we would appreciate citations to the project.

  author = {V. Fomin and J. Anmol and S. Desroziers and J. Kriss and A. Tejani},
  title = {High-level library to help with training neural networks in PyTorch},
  year = {2020},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{}},


We gratefully acknowledge the contributions of all open-source developers who helped get this project off of the ground and supported the project.

Special thanks and shout-out to Jeff Yang @ydcjeff for creating this beatiful website for the project!

Great work was done by Jeff Yang @ydcjeff and Taras Savchyn @trsvchn on PyTorch-Ignite Code-Generator app, thanks a lot!

We especially appreciate amazing work of Isabela Presedo-Floyd @isabela-pf on project’s logo, its elaboration and guidelines. Our new logos can be found here.

We are also grateful to Soumith Chintala, Joe Spisak from Meta for their help and to PyTorch org for infrastructure support (GPU CI and hosting of conda releases and our documentation).

In particular, Victor Fomin @vfdev-5 acknowledges Magellium and Quansight for supporting project’s development. Sylvain Desroziers @sdesrozis also gratefully acknowledges the support of IFPEN.

We also acknowledge scikit-learn project’s documentation on governance, author’s presentation etc from which this project is highly inspired.