Fetching the paper…

Looking Inward: Language Models Can Learn About Themselves by Introspection · Around