Hello! We’re having our first meeting of the year tomorrow, Thursday Jan 4th. We’re meeting on google meet.
PyTorch for Fedora
Thursday, January 4 · 9:00 – 9:30am
Time zone: America/New_York
Google Meet joining info
Video call link: https://meet.google.com/pfz-arqc-rue
Or dial: (GT) +502 2458 1186 PIN: 938 297 358 4693#
More phone numbers: https://tel.meet/pfz-arqc-rue?pin=9382973584693
Or join via SIP: sip:9382973584693@gmeet.redhat.com
Here are the minutes from today’s meeting.
How do we handle things like torchvision and huggingface WRT legal. Especially around downloaded weights. Is there a significant difference between including weights, automatically downloading weights and having to download weights by hand?
This will end up effecting llama-cpp, torchvision, huggingface, stable-diffusion.cpp and potentially others but llama-cpp and stable-diffusion.cpp are the least likely to be affected due to their requirements for downloading weights by hand.
Conclusion: leave questionable bits out of packages for now, tflink to start conversation with legal on what we’re allowed to include in the packages.
Still waiting on a few reviews, should be blocking the ML SIG review tracker
Getting close to enabling rocm based accelleration in pytorch but that’s still blocked on a few reviews (at least rocfft, hipfft once rocfft is reviewed).