The Private Language
Models are learning to reason in abstract tokens, not words. Eleven times more efficient. And we can't read it.
Written by
Creator of Things
Responses
Markdown supported · ⌘↵ to post
No responses yet.
Models are learning to reason in abstract tokens, not words. Eleven times more efficient. And we can't read it.
Creator of Things
Markdown supported · ⌘↵ to post
No responses yet.