Calibrating Verbal Uncertainty as a Linear Feature to Reduce Hallucinations
Authors:
- Ziwei JiZJZiwei Ji
- Lei YuLYLei Yu
- Yeskendir KoishekenovYKYeskendir Koishekenov
- Anthony HartshornAHAnthony Hartshorn
- Alan ScheltenASAlan Schelten
- Cheng ZhangCZCheng Zhang
- Nicola CanceddaNCNicola Cancedda
Top Cited Papers
Open Access
- 1 January 2025
- conference paper
- Published by Association for Computational Linguistics (ACL)
- p. 3769-3793
Abstract
Ziwei Ji, Lei Yu, Yeskendir Koishekenov, Yejin Bang, Anthony Hartshorn, Alan Schelten, Cheng Zhang, Pascale Fung, Nicola Cancedda. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.Keywords
All Related Versions
This publication has 31 references indexed in Scilit:
- On Early Detection of Hallucinations in Factual Question Answering2024
- Detecting hallucinations in large language models using semantic entropyNature, 2024
- "I'm Not Sure, But...": Examining the Impact of Large Language Models' Uncertainty Expression on User Reliance and Trust2024
- Survey of Hallucination in Natural Language GenerationACM Computing Surveys, 2023
- Enhancing Uncertainty-Based Hallucination Detection with Stronger Focus2023
- Navigating the Grey Area: How Expressions of Uncertainty and Overconfidence Affect Language Models2023
- When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories2023
- Reducing Conversational Agents’ Overconfidence Through Linguistic CalibrationTransactions of the Association for Computational Linguistics, 2022
- Controlling hallucinations at word level in data-to-text generationData Mining and Knowledge Discovery, 2021
- Natural Questions: A Benchmark for Question Answering ResearchTransactions of the Association for Computational Linguistics, 2019