Knowledge Verification to Nip Hallucination in the Bud

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wan, Fanqi, Huang, Xinting, Cui, Leyang, Quan, Xiaojun, Bi, Wei, Shi, Shuming
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913510935494656
author Wan, Fanqi
Huang, Xinting
Cui, Leyang
Quan, Xiaojun
Bi, Wei
Shi, Shuming
author_facet Wan, Fanqi
Huang, Xinting
Cui, Leyang
Quan, Xiaojun
Bi, Wei
Shi, Shuming
contents While large language models (LLMs) have demonstrated exceptional performance across various tasks following human alignment, they may still generate responses that sound plausible but contradict factual knowledge, a phenomenon known as hallucination. In this paper, we demonstrate the feasibility of mitigating hallucinations by verifying and minimizing the inconsistency between external knowledge present in the alignment data and the intrinsic knowledge embedded within foundation LLMs. Specifically, we propose a novel approach called Knowledge Consistent Alignment (KCA), which employs a well-aligned LLM to automatically formulate assessments based on external knowledge to evaluate the knowledge boundaries of foundation LLMs. To address knowledge inconsistencies in the alignment data, KCA implements several specific strategies to deal with these data instances. We demonstrate the superior efficacy of KCA in reducing hallucinations across six benchmarks, utilizing foundation LLMs of varying backbones and scales. This confirms the effectiveness of mitigating hallucinations by reducing knowledge inconsistency. Our code, model weights, and data are openly accessible at \url{https://github.com/fanqiwan/KCA}.
format Preprint
id arxiv_https___arxiv_org_abs_2401_10768
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Knowledge Verification to Nip Hallucination in the Bud
Wan, Fanqi
Huang, Xinting
Cui, Leyang
Quan, Xiaojun
Bi, Wei
Shi, Shuming
Computation and Language
While large language models (LLMs) have demonstrated exceptional performance across various tasks following human alignment, they may still generate responses that sound plausible but contradict factual knowledge, a phenomenon known as hallucination. In this paper, we demonstrate the feasibility of mitigating hallucinations by verifying and minimizing the inconsistency between external knowledge present in the alignment data and the intrinsic knowledge embedded within foundation LLMs. Specifically, we propose a novel approach called Knowledge Consistent Alignment (KCA), which employs a well-aligned LLM to automatically formulate assessments based on external knowledge to evaluate the knowledge boundaries of foundation LLMs. To address knowledge inconsistencies in the alignment data, KCA implements several specific strategies to deal with these data instances. We demonstrate the superior efficacy of KCA in reducing hallucinations across six benchmarks, utilizing foundation LLMs of varying backbones and scales. This confirms the effectiveness of mitigating hallucinations by reducing knowledge inconsistency. Our code, model weights, and data are openly accessible at \url{https://github.com/fanqiwan/KCA}.
title Knowledge Verification to Nip Hallucination in the Bud
topic Computation and Language
url https://arxiv.org/abs/2401.10768