Pretrain, finetune ANY AI model of ANY size on 1 or 10,000+ GPUs with zero code changes.
Fix calculate training time by summing all elapsed times instead the last one (#21291)
* fix: calculating training time by summing all differences instead of taking the last calculation * test timings * changelog --------- Co-authored-by: itzhaks <itzhak.stern@mobileye.com> Co-authored-by: Nicki Skafte Detlefsen <skaftenicki@gmail.com> Co-authored-by: Justus Schock <12886177+justusschock@users.noreply.github.com> Co-authored-by: Jirka Borovec <6035284+Borda@users.noreply.github.com>
I
itzhakstern committed
b554e9915aa9868bf62ae8feb2df0543272c1b3f
Parent: dd7b2f3
Committed by GitHub <noreply@github.com>
on 10/24/2025, 4:17:03 AM