LLMs can't self-correct in reasoning tasks, DeepMind study finds | Hacker News

#uncategorized   #link