Do Fedora developers plan to support AI development on Fedora?

Do Fedora developers plan to support AI development on Fedora? I search DNF for popular libraries and most of them are not found there.

There’s no such thing as “Fedora developers”. Fedora community members are volunteers who work on software that they use, mostly in their free time. At the moment, no community members are actively involved in an AI initiative (also mostly because a lot of AI uses GPUs which rely on proprietary software that will not be included in Fedora).

There is some interest in machine learning. Take a look at this thread:

https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/message/R5RATRDTJDEF7ZJKUMQSR2QTKY2O7W3D/

1 Like

As far as current GPUs and drivers, the proprietary software issue really only covers nvidia GPUs. AMD and Intel both have decent open source support in their drivers and (at least for AMD) for libraries like OpenCL/OpenMP. They just don’t have tensorflow and other libraries/hardware designed specifically to accelerate ML workloads.

The biggest issue, I would imagine, is probably finding people to maintain such packages an projects.

Another option would be to design an AI framework that’s capable of utilizing hybrid cluster/grid computing. MPI+OpenCL seems like a good candidate. That way, instead of having to use hardware that relies on proprietary software, people could just throw more computers at the problem. But that would be less efficient and would be a lot of work to design.

I would like to see more work on machine learning in the Fedora community though.
If I get more time to play around with this stuff, I’ll be sure to contribute where I can. For now, it’ll have to be someone else’s job.