Robust Uncertainty Quantification for Factual Generation of Large Language Models
面向大型语言模型事实生成的鲁棒不确定性量化
机构 * School of Cyberspace Security, Beijing University of Posts and Telecommunications(网络安全学院,北京邮电大学)
专题命中 知识编辑与模型理解 :large language model(title,abstract);language model(title,abstract);LLM(abstract);prompting(abstract)
AI总结 本文提出了一种新颖的鲁棒不确定性量化方法,用于评估大型语言模型在生成多事实任务中的可靠性,通过构建陷阱问题集验证了其有效性,提升了对抗性提问下的模型鲁棒性。
Comments 9 pages, 5 tables, 5 figures, accepted to IJCNN 2025
Journal ref 2025 International Joint Conference on Neural Networks (IJCNN), Rome, Italy, 2025, pp. 1-9