logoalt Hacker News

throwaway384034 • yesterday at 7:53 PM • 2 replies • view on HN

Sounds really bad for latency. Seems better to reason in the abstract or produce a mathematical model.


Replies

WarmWash • yesterday at 9:37 PM

Once a model gets good enough, etch it directly onto silicon and get 50,000 tokens per second.

largbae • yesterday at 7:54 PM

Aye basically use it to make synthetic training data.

➕ show 1 reply