Pretrain, finetune ANY AI model of ANY size on 1 or 10,000+ GPUs with zero code changes.
Fix warning frequency in throughput monitor callback (#21453)
* only warn once in ThroughputMonitor * add test * renaming * chlog
G
GdoongMathew committed
aa0ee0d18d49c6b26f18d34a3473f177adefc262
Parent: 9df1910
Committed by GitHub <[email protected]>
on 1/14/2026, 9:33:45 AM