LLMs think in terms of language, and they predictably seem very wrapped up in the abstractions we've built language out of. they're not very good at recognizing when those abstractions aren't…

Posters: xsphi. xsphi

4 likes recorded by the source corpus.