News for hackers. Stuff that matters.

A comment on Internal representations of LLMs encode information about truthfulness. Parent comment

Re:Internal representations of LLMs encode information about... by ben_w on Wednesday October 30, 2024 @11:39AM (#41996318)
Even monolithic training runs take sources more disparate than any human has the capacity to consume.

Also, given the lack of imagination everyone has with naming places, I had to check:

https://en.wikipedia.org/wiki/Moscow_(disambiguation)

[ Parent | on HN ]
Re:Internal representations of LLMs encode information about... by stoniejohnson on Wednesday October 30, 2024 @01:16PM (#41997726)
I was responding to the idea that an LLM would believe (regurgitate) untrue things if you pretrained them on untrue things. I wasn't making a claim about SOTA models with gigantic training corpora.
[ Parent | on HN ]