LLMs cannot find reasoning errors, but can correct them by from Hacker News on 2023-11-20 19:35 (#6GH9K) Comments