PyTorch and Microsoft Azure logo

PyTorch is an increasingly popular open-source deep learning framework that accelerates AI innovations from research to production. At Microsoft, we use PyTorch to power products such as Bing and Azure Cognitive Services and we actively contribute to several PyTorch open-source projects, including PyTorch Profiler, ONNX Runtime, DeepSpeed, and more.

Today, we’re announcing a new initiative in collaboration with Facebook—the PyTorch Enterprise Support Program. This new program enables service providers to develop and offer tailored enterprise-grade support to their customers.

As one of the founding members of the PyTorch Enterprise Support Program, Microsoft is launching PyTorch Enterprise on Azure to provide long-term support, prioritized troubleshooting, and integration with Azure solutions. Azure is the first cloud to provide enterprise support for PyTorch and the most performant and reliable place for enterprise workloads.

 

Nuance logo, company name“This new enterprise-level offering by Microsoft closes an important gap. PyTorch gives our researchers unprecedented flexibility in designing their models and running their experiments. Serving these models in production, however, can be a challenge. The direct involvement of Microsoft lets us deploy new versions of PyTorch to Azure with confidence.”—Jeremy Jancsary, Senior Principal Research Scientist, Nuance

Provide commercial support to PyTorch

It is easy to get started and build a proof of concept with PyTorch. However, PyTorch users can run into challenges and may want someone to help when using PyTorch in production. Conversations with customers led to the inception of PyTorch Enterprise which addresses these issues by providing:

  • Long-term support (LTS): Microsoft will provide commercial support for the public PyTorch codebase. Each release will be supported for as long as it is current. In addition, one PyTorch release will be selected for LTS every year. Such releases will be supported for two years, enabling a stable production experience without frequent major upgrade investment.
  • Prioritized troubleshooting: Microsoft Enterprise support customers, including Premier and Unified, are automatically eligible for PyTorch Enterprise at no additional cost. The dedicated PyTorch team in Azure will prioritize, develop, and deliver hotfixes to customers as needed. These hotfixes will get tested and will be included in future PyTorch releases. In addition, Microsoft will extensively test PyTorch releases for performance regressions with continuous integration and realistic, demanding workloads from internal Microsoft applications.
  • Azure integration: The latest release of PyTorch will be integrated with Azure Machine Learning, along with other PyTorch add-ons, including ONNX Runtime for faster inferencing. Microsoft will continue to invest in the ONNX standard to improve PyTorch inference and training speed.

Crayon logo“Scaling up machine learning models to enterprise-level AI applications is a challenge Crayon addresses for our customers every day. We have been using PyTorch on Azure and enjoying the smooth integration. With PyTorch Enterprise, we have more confidence to leverage the most cutting-edge features offered by newer PyTorch versions in our customers’ projects.”—Tailai Wen, Lead Data Scientist, Crayon

Contribute back to open source

PyTorch Enterprise benefits not only Azure customers but also the PyTorch community users. Selected code that aligns with PyTorch will be fed back to the public PyTorch distribution so everyone in the community can benefit.

NVIDIA logo“NVIDIA works closely with the PyTorch community to ensure the best user experience on NVIDIA GPUs. PyTorch Enterprise from Microsoft Azure will take the experience of building, deploying, and managing AI applications a step further, enabling data scientists to get the most from their NVIDIA investment.”—Christian Sarofeen, Senior Engineering Manager, PyTorch Frameworks Team, NVIDIA

Configurations supported

PyTorch Enterprise supports the following configurations:

  • PyTorch: version 1.8.1 and up.
  • Libraries: torch, torchaudio, torchvision, torchtext, onnxruntime, and torch-tb-profiler.
  • Python: version 3.6 and up.
  • NVIDIA CUDA: versions 10.2 and 11.1.
  • Operating systems: Windows 10, Debian 9, Debian 10, Ubuntu 16.04.7 LTS, and Ubuntu 18.04.5 LTS (x86_64 architecture only).

PyTorch Enterprise is currently available on Azure Machine Learning and Data Science Virtual Machines (DSVM). It will soon be available on Azure Synapse Analytics.

PyTorch Enterprise does not currently support:

Get started with PyTorch Enterprise

To get started with PyTorch Enterprise, join the Microsoft Premier or Unified support program. Contact your Microsoft account representative for additional information on different enterprise support options.

If you would like to try out the PyTorch LTS version, you can do so at PyTorch.org.

Learn more