‹ BackHN Continuity

Thread

Show HN: TurboGPT: train 22KiB transformer in 13s

56 points · 10 comments · lostmsu

  1. jey · · focus · HN ↗
    At that scale it seems like you could just solve the KKT conditions directly. Exaggerating but only a little
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.