HalluLens: LLM Hallucination Benchmark

2025-04-24

AI Summary

LLMs’ “hallucinations”—factually incorrect outputs contradicting either input prompts or training data—severely undermine their trustworthiness and practical applicability. Existing work suffers from ambiguous definitions, inconsistent taxonomies, and saturated or data-leaking evaluation benchmarks. To address these limitations, we propose the first comprehensive hallucination benchmark for LLMs. Our method introduces three core innovations: (1) a novel decoupling of “hallucination” from “factualness,” establishing a principled dichotomy between *exogenous* hallucinations (violating training knowledge) and *endogenous* hallucinations (violating user-provided input); (2) a dynamic test-set generation mechanism to prevent data contamination; and (3) a multi-dimensional consistency evaluation framework, coupled with a systematic critique of prevailing benchmarks. The open-sourced benchmark uncovers deep-rooted issues—including factual confusion—and provides a standardized, infrastructure-level assessment platform for advancing LLM reliability research.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Large language models (LLMs) often generate responses that deviate from user input or training data, a phenomenon known as"hallucination."These hallucinations undermine user trust and hinder the adoption of generative AI systems. Addressing hallucinations is essential for the advancement of LLMs. This paper introduces a comprehensive hallucination benchmark, incorporating both new extrinsic and existing intrinsic evaluation tasks, built upon clear taxonomy of hallucination. A major challenge in benchmarking hallucinations is the lack of a unified framework due to inconsistent definitions and categorizations. We disentangle LLM hallucination from"factuality,"proposing a clear taxonomy that distinguishes between extrinsic and intrinsic hallucinations, to promote consistency and facilitate research. Extrinsic hallucinations, where the generated content is not consistent with the training data, are increasingly important as LLMs evolve. Our benchmark includes dynamic test set generation to mitigate data leakage and ensure robustness against such leakage. We also analyze existing benchmarks, highlighting their limitations and saturation. The work aims to: (1) establish a clear taxonomy of hallucinations, (2) introduce new extrinsic hallucination tasks, with data that can be dynamically regenerated to prevent saturation by leakage, (3) provide a comprehensive analysis of existing benchmarks, distinguishing them from factuality evaluations.

Problem

• Define clear taxonomy for LLM hallucination types
• Develop dynamic benchmark to prevent data leakage
• Distinguish hallucination from factuality in LLM outputs

Innovation

• Introduces clear hallucination taxonomy framework
• Dynamic test set prevents data leakage
• Distinguishes extrinsic intrinsic hallucinations