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.
@article{arxiv.2504.21773,
title = {MAC-Tuning: LLM Multi-Compositional Problem Reasoning with Enhanced Knowledge Boundary Awareness},
author = {Junsheng Huang and Zhitao He and Yucheng Huang and Sandeep Polisetty and Qingyun Wang and Yi. R Fung},
journal= {arXiv preprint arXiv:2504.21773},
year = {2025}
}
Comments
We release our code and resource at https://github.com/no-touch-fish/Multi-QA-Tuning. The paper is accepted into EMNLP 2025 main