Pretrain, finetune ANY AI model of ANY size on 1 or 10,000+ GPUs with zero code changes.
Cast on host instead of IPU when using `precision=16` (#13880)
Co-authored-by: Carlos Mocholí <[email protected]> Co-authored-by: Rohit Gupta <[email protected]>
H
HMellor committed
07b39c257bac44d82a5d704d79ef103c829597a2
Parent: 233b36b
Committed by GitHub <[email protected]>
on 7/28/2022, 7:26:41 PM