Microsoft Launches PyTorch Enterprise on Azure

Today, Microsoft unveiled PyTorch Enterprise, a new service within Azure designed to furnish developers with enhanced assistance when utilizing PyTorch on the Azure platform. Essentially, this represents Microsoft’s formalized commercial support package for the PyTorch framework.

Understanding PyTorch

PyTorch is an open-source machine learning framework primarily built around the Python programming language. Its strengths lie in the domains of computer vision and natural language processing. Originally created by Facebook, it offers functionality comparable to Google’s widely-used TensorFlow framework.

Frank X. Shaw, Microsoft’s Corporate Vice President of Communications, characterized the PyTorch Enterprise service as enabling developers to achieve “a more dependable production experience for organizations leveraging PyTorch in their data science initiatives.”

Benefits of PyTorch Enterprise

Subscribers to Microsoft’s Premier and Unified Support programs will gain access to several advantages through PyTorch Enterprise. These include prioritized support requests, direct hands-on assistance, and readily available solutions for hotfixes, bug resolutions, and security updates.

Furthermore, Microsoft commits to designating one PyTorch version annually for extended, long-term support.

Azure has already streamlined the process of using PyTorch. Microsoft’s investment in the library is demonstrated by its assumption of PyTorch development for Windows last year. As highlighted in today’s announcement, upcoming PyTorch releases will integrate seamlessly with Azure Machine Learning.

The company also pledges to contribute code derived from developer feedback back to the broader public PyTorch distribution.

Availability and Support

PyTorch Enterprise support will be available for versions 1.8.1 and later, running on both Windows 10 and a variety of prevalent Linux distributions.

Jeremy Jancsary, Senior Principal Research Scientist at Nuance, stated, “This new enterprise-level offering from Microsoft addresses a significant need. PyTorch provides our researchers with exceptional flexibility in model design and experimentation.”

He continued, “However, deploying these models into production environments can present challenges. Microsoft’s direct involvement allows us to confidently deploy new PyTorch versions to Azure.”

A Strategic Move

With this new service, Microsoft is adopting a common monetization strategy seen in the open-source world – providing premium services built upon an open-source project.

Given that PyTorch was not initially developed by a startup subsequently acquired by a cloud provider, this move is perceived as a relatively straightforward and logical extension of existing offerings.