Comment

Comment by nottorp

Of course, the initial mistake is assuming LLMs have any 'reasoning'...
parent nottorp 2 years ago

Discussion

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton