HN
Hacker News
Top
New
Best
Ask
Show
Jobs
Comment by cypress66 | Hacker News Reader
Parent
Full thread
cypress66
·
Gpt3 includes dense attention layers that are n^2
View on HN
axiom92
·
Sure, but we don't know if ChatGPT is based on the original GPT-3 architecture.
Reply on news.ycombinator.com