Re:LLMs can't self-correct in reasoning tasks, DeepMind stud...
by jncfhnb on Tuesday October 10, 2023 @09:09PM
(#37839806)
LLMs “think” in vector spaces, and “operate” by picking out words one at a time. A full “idea” is there from the beginning.
If you say “choose X or Y and then give three supporting arguments” the LLM does not write out the supporting arguments, but the vector space determining the initial one word answer of X or Y does include the embedded awareness of those arguments to different levels of specificity and relevance.
The annoying disclaimers and refusals are not fundamental to LLMs but specific LLM services.
If you say “choose X or Y and then give three supporting arguments” the LLM does not write out the supporting arguments, but the vector space determining the initial one word answer of X or Y does include the embedded awareness of those arguments to different levels of specificity and relevance.
The annoying disclaimers and refusals are not fundamental to LLMs but specific LLM services.