Pretrain, finetune ANY AI model of ANY size on 1 or 10,000+ GPUs with zero code changes.
[TPU] Set the compute_dtype with XLAFSDP (#18497)
C
Carlos Mocholí committed
e1c5c5ae4aad7916c531eaf2ff60138c56b4c4ca
Parent: cc18bd7
Committed by GitHub <[email protected]>
on 9/7/2023, 4:43:21 PM