
In 2016, Meta (then Facebook) introduced the Pytorch framework, an open-source AI research library. After six years and 150,000 projects from 2,400 people, Meta announced that the Pytorch project would soon be spun out from the company’s direct supervision to become its own institution, the Pytorch Foundation, a subsidiary inside the wider Linux Foundation nonprofit umbrella.
Pytorch has developed to become a prominent standard for the AI research community over the last half decade, with Meta CEO Mark Zuckerberg noting in a news statement on Monday that about 80% of “researchers who submit their work at major ML conferences, such as NeurIPS or ICML, harness the framework.”
“We have built libraries that support some of the principal domains of the AI field, such as torchvision, which powers most of the world’s modern computer vision research,” Zuckerberg added. “The framework will continue to be a part of Meta’s AI research and engineering work.”
However, Pytorch is more than just Meta’s baby; it also serves as a technological foundation for most of Amazon’s Web Services effort, as well as Microsoft Azure and OpenAI. As a result, the Pytorch Foundation “will boast a wide-ranging governing board composed of representatives from AMD, Amazon Web Services, Google Cloud, Meta, Microsoft Azure, and Nvidia, with the intention to expand further over time.” To guarantee that the new organization does not lose sight of the ideals it represents, it will follow four principles: “remaining open, maintaining neutral branding, staying fair, and forging a strong technical identity.” “Don’t be evil” appears to have already been taken.
Despite the fact that it is no longer subject to direct management, Meta aims to continue using Pytorch as its principal AI research platform and to financially support it accordingly. However, Zuckerberg stated that the company will preserve “a clear separation between the business and technical governance” of the foundation.