undefined | Better HN

0 pointsingenieroariel3y ago0 comments

Try switching all the .cuda() to .mps() I got a 100x speedup on a different language model on a Macbook M1 Air.

0 comments

1 comments · 1 top-level

j / k navigate · click thread line to collapse