Fedora 40 Looks At Packaging Its Own PyTorch

While on Fedora and other Linux distributions it can be as easy as running "pip3 install torch" or similar for deploying the PyTorch machine learning framework, Fedora 40 is looking at packaging PyTorch on its own for enhancing the Fedora Linux user experience.

Similar to select Python packages being packaged on their own by Fedora Linux to complement what's already available from PyPi, Red Hat engineer Tom Rix is proposing that Fedora 40 package its own PyTorch.

The goal of Fedora packaging PyTorch on its own is for ensuring this machine learning framework is more easily accessible and "seamlessly integrable" within the Fedora Linux ecosystem. Fedora 40 would be the first time PyTorch is packaged up by this Linux distribution. They would be focusing on PyTorch for CPU acceleration on x86_64 and AArch64 as well as ongoing development efforts around PyTorch for AMD GPUs.

The change proposal for Fedora 40 argues that with the Pip-based installation building locally on the system, there is chances it could fail and not be consistent from machine-to-machine. Thus the hope is that telling users to sudo dnf install python-torch will be more consistent and easier.

See the Fedora Wiki for those interested in this Fedora 40 change proposal that still needs to be voted on by the Fedora Engineering and Steering Committee (FESCo). The stable Fedora 40 release is due out in April.