arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2404.06948cs.CLcs.AI

MetaCheckGPT——一种利用大语言模型不确定性与元模型的多任务幻觉检测器

MetaCheckGPT -- A Multi-task Hallucination Detector Using LLM Uncertainty and Meta-models

IIIT Hyderabad(印度理工学院海得拉巴分校);Purdue University(普渡大学);Northwestern University(西北大学);Indiana University Indianapolis(印第安纳大学印第安纳波利斯分校)

Rahul Mehta, Andrew Hoblitzell, Jack O'Keefe, Hyeju Jang, Vasudeva Varma

AI总结:

针对大语言模型的幻觉问题,本文提出基于LLM元回归器的MetaCheckGPT检测框架,在Semeval 2024 SHROOM任务的两个子赛道分获冠亚军,还通过实验对比揭示了GPT-4的局限性。

摘要:

大语言模型(LLM)的幻觉问题近来已成为一个重要难题。该方向的一项最新研究是Semeval 2024 Task 6的共享任务SHROOM(幻觉及相关可观测过度生成错误共享任务)。本文介绍了我们的获奖方案,该方案在模型无关和模型感知两个子任务赛道中分别位列第1名和第2名。我们提出了一种用于模型评估与集成的LLM元回归器框架,在排行榜上取得了最高分。我们还对多种基于transformer的模型以及ChatGPT、Vectara等黑盒方法进行了实验。此外,我们将GPT-4与我们的最优模型进行对比并开展了误差分析,揭示了前者的局限性。

英文摘要:

Hallucinations in large language models (LLMs) have recently become a significant problem. A recent effort in this direction is a shared task at Semeval 2024 Task 6, SHROOM, a Shared-task on Hallucinations and Related Observable Overgeneration Mistakes. This paper describes our winning solution ranked 1st and 2nd in the 2 sub-tasks of model agnostic and model aware tracks respectively. We propose a meta-regressor framework of LLMs for model evaluation and integration that achieves the highest scores on the leaderboard. We also experiment with various transformer-based models and black box methods like ChatGPT, Vectara, and others. In addition, we perform an error analysis comparing GPT4 against our best model which shows the limitations of the former.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑