Enregistré dans:
Détails bibliographiques
Auteurs principaux: Paudel, Bibek, Lyzhov, Alexander, Joshi, Preetam, Anand, Puneet
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2504.07069
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866910907567702016
author Paudel, Bibek
Lyzhov, Alexander
Joshi, Preetam
Anand, Puneet
author_facet Paudel, Bibek
Lyzhov, Alexander
Joshi, Preetam
Anand, Puneet
contents This paper introduces a comprehensive system for detecting hallucinations in large language model (LLM) outputs in enterprise settings. We present a novel taxonomy of LLM responses specific to hallucination in enterprise applications, categorizing them into context-based, common knowledge, enterprise-specific, and innocuous statements. Our hallucination detection model HDM-2 validates LLM responses with respect to both context and generally known facts (common knowledge). It provides both hallucination scores and word-level annotations, enabling precise identification of problematic content. To evaluate it on context-based and common-knowledge hallucinations, we introduce a new dataset HDMBench. Experimental results demonstrate that HDM-2 out-performs existing approaches across RagTruth, TruthfulQA, and HDMBench datasets. This work addresses the specific challenges of enterprise deployment, including computational efficiency, domain specialization, and fine-grained error identification. Our evaluation dataset, model weights, and inference code are publicly available.
format Preprint
id arxiv_https___arxiv_org_abs_2504_07069
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle HalluciNot: Hallucination Detection Through Context and Common Knowledge Verification
Paudel, Bibek
Lyzhov, Alexander
Joshi, Preetam
Anand, Puneet
Computation and Language
Artificial Intelligence
This paper introduces a comprehensive system for detecting hallucinations in large language model (LLM) outputs in enterprise settings. We present a novel taxonomy of LLM responses specific to hallucination in enterprise applications, categorizing them into context-based, common knowledge, enterprise-specific, and innocuous statements. Our hallucination detection model HDM-2 validates LLM responses with respect to both context and generally known facts (common knowledge). It provides both hallucination scores and word-level annotations, enabling precise identification of problematic content. To evaluate it on context-based and common-knowledge hallucinations, we introduce a new dataset HDMBench. Experimental results demonstrate that HDM-2 out-performs existing approaches across RagTruth, TruthfulQA, and HDMBench datasets. This work addresses the specific challenges of enterprise deployment, including computational efficiency, domain specialization, and fine-grained error identification. Our evaluation dataset, model weights, and inference code are publicly available.
title HalluciNot: Hallucination Detection Through Context and Common Knowledge Verification
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2504.07069