Looking Inward: Language Models Can Learn About Themselves by Introspection

Open in new window