1 paper · 1 filter
Dhruv Madhwal, Lyuxin David Zhang, Dan Roth +2
Large language models often struggle to recognize their knowledge limits in closed-book question answering, leading to confident hallucinations. While decomposed prompting is typic…