PyTorch Enterprise Support Program is a participatory program that enables service providers to develop and offer tailored enterprise-grade support to their customers. This program was created in direct response to feedback from PyTorch enterprise users who are deploying models in production at scale for mission-critical applications.
Read the program launch blog post.
Support Services by Participants
Microsoft is one of the founding members and also the inaugural participant of the program. Microsoft offers PyTorch Enterprise on Azure as a part of Microsoft Premier and Unified Support. The PyTorch Enterprise support service includes long-term support to selected versions of PyTorch for up to 2 years, prioritized troubleshooting, and the latest integration with Azure Machine Learning and other PyTorch add-ons including ONNX Runtime for faster inference.
To learn more and get started with PyTorch Enterprise on Azure, visit here.
For documentation, visit here.
For nightly tests, visit here.
OPEN SOURCE CONTRIBUTIONS
All hotfixes done by Participants for their enterprise customers will be tested and fed back to the PyTorch long-term support (LTS) and other branches regularly. As a PyTorch user, you can use the LTS release branches of PyTorch. Program Participants will prioritize and answer pull requests of customers subscribed to their support services. Any other PRs are reviewed the same as how they are handled for PyTorch. Development of PyTorch, including LTS branches and releases as a foundation for the PyTorch Enterprise Support Program, will continue to be in the open and open source.
As a Participant of the program, providers are able to:
- Provide a high-value 24x7 support service including troubleshooting and hotfixes for customers running PyTorch in mission-critical production environments.
- Work with other Participants to share and improve PyTorch LTS point releases, including contributions of hotfixes and other improvements found while working closely with customers and on their systems.
- Add test coverage for these fixes when PyTorch LTS releases are created.
- Create an enterprise support service and brand it using the PyTorch trademark and logo, following the terms and conditions set forth in the PyTorch Enterprise Support Program.