arXiv 2504.217732025-09-14cs.CLcs.AI

MAC-Tuning：具有增强知识边界感知的LLM多组合问题推理

MAC-Tuning: LLM Multi-Compositional Problem Reasoning with Enhanced Knowledge Boundary Awareness

Hong Kong University of Science and Technology（香港科技大学）；University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；UMass Amherst（马萨诸塞大学阿姆赫斯特分校）；William & Mary（威廉与玛丽学院）

Junsheng Huang, Zhitao He, Yucheng Huang, Sandeep Polisetty, Qingyun Wang, Yi. R Fung

备注：: We release our code and resource at https://github.com/no-touch-fish/Multi-QA-Tuning. The paper is accepted into EMNLP 2025 main

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对LLM在多问题设定下捏造事实的幻觉问题，提出MAC-Tuning方法，在指令微调时分离答案预测与置信度估计的学习，实验表明其平均精度较基线提升达25%。

摘要：

鉴于LLM在各种应用中的广泛采用，其捏造不存在事实的幻觉问题显得尤为重要。以往研究通过分析内部参数化知识边界来估计置信度以解决此问题。然而，这些研究集中于单问题设定，尚未探索更具挑战性的多问题设定，该设定要求同时准确回答多个问题。我们提出了一种针对多问题设定的新方法，即多答案与置信度逐步调整（MAC-Tuning），该方法在指令数据微调期间将答案预测和置信度估计的学习分离开来。大量实验表明，我们的方法在平均精度上比基线高出25%。

英文摘要：

The hallucination of non-existent facts by LLMs is an important problem given its widespread adoption across various applications. Previous research addresses this problem by analyzing the internal parameterized knowledge boundaries to estimate confidence. However, these studies focus on the single-problem setting and have not explored the more challenging multi-problem setting, which requires accurately answering multiple questions simultaneously. We introduce a novel method for the multi-problem setting, Multiple Answers and Confidence Stepwise Tuning (MAC-Tuning), that separates the learning of answer prediction and confidence estimation during fine-tuning on instruction data. Extensive experiments demonstrate that our method outperforms baselines by up to 25\% in average precision.

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。