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 professor_v on Wednesday October 30, 2024 @12:34PM (#41997134)
I'm extremely skeptical about this, I once believed the internet would do something similar and it seems to have done exactly the opposite.
[ Parent | on HN ]
Re:Internal representations of LLMs encode information about... by topspin on Wednesday October 30, 2024 @12:51PM (#41997358)
Indeed. A model is only as good as its data. Propagandists have no difficulty grooming inputs. We have already seen high profile cases of this with machine learning.
[ Parent | on HN ]
Re:Internal representations of LLMs encode information about... by mdp2021 on Wednesday October 30, 2024 @01:42PM (#41998053)
Completely different things.

"Will the ability to let everyone express increase noise?" // Yes

"Will feeding all available data to a processor reduce noise?" // Probably

[ Parent | on HN ]