LLM Internal States Reveal Hallucination Risk Faced With a Query — mentioned by redlock

Sat, 15 Feb 2025 12:50:10 GMT

Parent poster did say they are aware they don’t know but can’t express it. I am guessing he is referring to mechanistic interpretability research like these: https://arxiv.org/abs/2405.16908 https://arxiv.org/abs/2407.03282 You are claiming they are statistical parrots, which I don’t think the parent poster meant. The “statistical parrots” argument might have been compelling with GPT-3, but not with today’s models and the results of mechanistic interpretability research, which show internal representations and rudimentary world models.

Paper and discussion history. A mention is not an endorsement.

LLM Internal States Reveal Hallucination Risk Faced With a Query · HackerPapers discussions

LLM Internal States Reveal Hallucination Risk Faced With a Query — mentioned by redlock