1 paper · 1 filter
Yangfan Hu, Xuhan Tong, Haoyue Bai +5
Large language models often produce hallucinated answers that violate prompt-level constraints. A key diagnostic question is whether these failures reflect missing knowledge, or wh…