作者位次 #5/8ACL 2026Poster· Natural Language Processing
Beyond " I Don’t Know": Evaluating LLM Self-Awareness in Discriminating Data and Model Uncertainty
Jingyi Ren · Ante Wang · Yunghwei Lai · Xiaolong Wang · Linlu Gong · Weitao Li · Weizhi Ma · Yang Liu
摘要
Reliable Large Language Models (LLMs) should abstain when confidence is insufficient. However, prior studies often treat refusal as a generic "I don’t know”, failing to distinguish input-level ambiguity (data uncertainty) from capability limitations (model uncertainty). This lack of distinction l...