1
SeSE: Black-Box Uncertainty Quantification for Large Language Models Based on Structural Information Theory
基于结构信息理论的黑盒不确定性量化方法，无需访问模型内部，已被UAI 2026接收，为LLM可靠性评估提供新思路。
arXiv:2511.16275v4 Announce Type: replace-cross Abstract: Reliable uncertainty quantification (UQ) is essential for deploying large language models (L…