logoalt Hacker News

codeliontoday at 4:50 AM0 repliesview on HN

How does it compare to some of the newer mlx inference engines like optiq that support turboquantization - https://mlx-optiq.pages.dev/