← Back to context

Comment by p1esk

1 day ago

They experimented with gpt-2 scale models. Hard to make any meaningful conclusions in the gpt-5 era.

0 comments

p1esk

Reply

No comments yet

Contribute on Hacker News ↗