@inproceedings{kadotani-etal-2025-learning,
title = "Learning from Hallucinations: Mitigating Hallucinations in {LLM}s via Internal Representation Intervention",
author = "Kadotani, Sora and
Nishida, Kosuke and
Nishida, Kyosuke",
editor = "Inui, Kentaro and
Sakti, Sakriani and
Wang, Haofen and
Wong, Derek F. and
Bhattacharyya, Pushpak and
Banerjee, Biplab and
Ekbal, Asif and
Chakraborty, Tanmoy and
Singh, Dhirendra Pratap",
booktitle = "Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics",
month = dec,
year = "2025",
address = "Mumbai, India",
publisher = "The Asian Federation of Natural Language Processing and The Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-ijcnlp.68/",
pages = "1136--1143",
ISBN = "979-8-89176-303-6",
abstract = "Large language models (LLMs) sometimes hallucinate facts. Recent studies have shown that use of non-factual LLMs (anti-expert) have the potential to improve the factuality of the base LLM. Anti-expert methods penalize the output probabilities of the base LLM with an anti-expert LLM. Anti-expert methods are effective in mitigating hallucinations, but require high computational costs because the two LLMs are run simultaneously. In this paper, we propose an efficient anti-expert method called in-model anti-expert. It mitigated the hallucination problem with a single LLM and intervening to change the internal representations in the direction of improving factuality. Experiments results showed that the proposed method is less costly than the conventional anti-expert method and outperformed existing methods except for the anti-expert method. We confirmed that the proposed method improved GPU memory usage from 2.2x to 1.2x and latency from 1.9x to 1.2x."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kadotani-etal-2025-learning">
<titleInfo>
<title>Learning from Hallucinations: Mitigating Hallucinations in LLMs via Internal Representation Intervention</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sora</namePart>
<namePart type="family">Kadotani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kosuke</namePart>
<namePart type="family">Nishida</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kyosuke</namePart>
<namePart type="family">Nishida</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kentaro</namePart>
<namePart type="family">Inui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sakriani</namePart>
<namePart type="family">Sakti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haofen</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Derek</namePart>
<namePart type="given">F</namePart>
<namePart type="family">Wong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pushpak</namePart>
<namePart type="family">Bhattacharyya</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Biplab</namePart>
<namePart type="family">Banerjee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Asif</namePart>
<namePart type="family">Ekbal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dhirendra</namePart>
<namePart type="given">Pratap</namePart>
<namePart type="family">Singh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>The Asian Federation of Natural Language Processing and The Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Mumbai, India</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-303-6</identifier>
</relatedItem>
<abstract>Large language models (LLMs) sometimes hallucinate facts. Recent studies have shown that use of non-factual LLMs (anti-expert) have the potential to improve the factuality of the base LLM. Anti-expert methods penalize the output probabilities of the base LLM with an anti-expert LLM. Anti-expert methods are effective in mitigating hallucinations, but require high computational costs because the two LLMs are run simultaneously. In this paper, we propose an efficient anti-expert method called in-model anti-expert. It mitigated the hallucination problem with a single LLM and intervening to change the internal representations in the direction of improving factuality. Experiments results showed that the proposed method is less costly than the conventional anti-expert method and outperformed existing methods except for the anti-expert method. We confirmed that the proposed method improved GPU memory usage from 2.2x to 1.2x and latency from 1.9x to 1.2x.</abstract>
<identifier type="citekey">kadotani-etal-2025-learning</identifier>
<location>
<url>https://aclanthology.org/2025.findings-ijcnlp.68/</url>
</location>
<part>
<date>2025-12</date>
<extent unit="page">
<start>1136</start>
<end>1143</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Learning from Hallucinations: Mitigating Hallucinations in LLMs via Internal Representation Intervention
%A Kadotani, Sora
%A Nishida, Kosuke
%A Nishida, Kyosuke
%Y Inui, Kentaro
%Y Sakti, Sakriani
%Y Wang, Haofen
%Y Wong, Derek F.
%Y Bhattacharyya, Pushpak
%Y Banerjee, Biplab
%Y Ekbal, Asif
%Y Chakraborty, Tanmoy
%Y Singh, Dhirendra Pratap
%S Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics
%D 2025
%8 December
%I The Asian Federation of Natural Language Processing and The Association for Computational Linguistics
%C Mumbai, India
%@ 979-8-89176-303-6
%F kadotani-etal-2025-learning
%X Large language models (LLMs) sometimes hallucinate facts. Recent studies have shown that use of non-factual LLMs (anti-expert) have the potential to improve the factuality of the base LLM. Anti-expert methods penalize the output probabilities of the base LLM with an anti-expert LLM. Anti-expert methods are effective in mitigating hallucinations, but require high computational costs because the two LLMs are run simultaneously. In this paper, we propose an efficient anti-expert method called in-model anti-expert. It mitigated the hallucination problem with a single LLM and intervening to change the internal representations in the direction of improving factuality. Experiments results showed that the proposed method is less costly than the conventional anti-expert method and outperformed existing methods except for the anti-expert method. We confirmed that the proposed method improved GPU memory usage from 2.2x to 1.2x and latency from 1.9x to 1.2x.
%U https://aclanthology.org/2025.findings-ijcnlp.68/
%P 1136-1143
Markdown (Informal)
[Learning from Hallucinations: Mitigating Hallucinations in LLMs via Internal Representation Intervention](https://aclanthology.org/2025.findings-ijcnlp.68/) (Kadotani et al., Findings 2025)
ACL