Re:LLM Hallucinations: why they occur and how to avoid them
by jamilton on Tuesday May 02, 2023 @10:41PM
(#35797111)
That framing fails to describe the case where the model is confident in a response, and is wrong, which I think is still considered hallucinating.
[ Parent
| on HN ]
Re:LLM Hallucinations: why they occur and how to avoid them
by ajcp on Tuesday May 02, 2023 @10:50PM
(#35797186)
How can a model return high-level confidence at a token level on data it can predict over.
Re:LLM Hallucinations: why they occur and how to avoid them
by jamilton on Wednesday May 03, 2023 @01:32AM
(#35798388)
Misconceptions. There's no inherent reason a false statement would have lower probability than a true one.
To be clear, I'm referring to things like GPT-3.5 reportedly consistently messing up on statements like "what's heavier, two pounds of feathers or a pound of bricks". Being consistently wrong in the same way implies to me (but I don't know for sure) that the class of response is high probability in an absolute sense.
I can't find the article that demonstrated the sort of things that GPT consistently gets wrong, but it was things like common misconceptions and sayings.
Re:LLM Hallucinations: why they occur and how to avoid them
by ajcp on Wednesday May 03, 2023 @03:28PM
(#35806609)
Very interesting. So it could produce, with high confidence, common and real-world guesses found in it's dataset.
So in that case it's not guessing and not wrong; it's indeed producing something that is correct, but still false. Now we're really getting into the weeds here though.