LLMs slurp up a lot of trolling and typical tech sarcasm through its training da... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		lynx97 82 days ago \| parent \| context \| favorite \| on: Yes-rs: A fast, memory-safe rewrite of the classic... LLMs slurp up a lot of trolling and typical tech sarcasm through its training data. IMO a reason for "hallucinations".

alpaca128 82 days ago [–]

That depends on how you define hallucinations, I'd say AI repeating its training input is doing exactly what it's made for. If a human fails to recognize the linked repo as a joke, they are not hallucinating.

lynx97 82 days ago | [–]

Thats why I put hallucinations in quotes.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact