RAGTruth: A Hallucination Corpus for Developing Trustworthy Retrieval-Augmented Language Models

By Yuanhao Wu, Juno Zhu, Siliang Xu, Kashun Shum, Cheng Niu, Randy Zhong, Juntong Song, Tong Zhang. Annual Meeting of the Association for Computational Linguistics, 2023

319 citations recorded by the source corpus.

Retrieval-augmented generation (RAG) has become a main technique for alleviating hallucinations in large language models (LLMs). Despite the integration of RAG, LLMs may still present unsupported or contradictory claims to the retrieved contents. In order to develop effective hallucination prevention strategies under RAG, it is important to create benchmark datasets that can measure the extent of hallucination. This paper presents RAGTruth, a corpus tailored for analyzing word-level hallucinations in various domains and tasks within the standard RAG frameworks for LLM applications. RAGTruth comprises nearly 18,000 naturally generated responses from diverse LLMs using RAG. These responses have undergone meticulous manual annotations at both the individual cases and word levels, incorporating evaluations of hallucination intensity. We not only benchmark hallucination frequencies across different LLMs, but also critically assess the effectiveness of several existing hallucination detection methodologies. Furthermore, we show that using a high-quality dataset such as RAGTruth, it is possible to finetune a relatively small LLM and achieve a competitive level of performance in hallucination detection when compared to the existing prompt-based approaches using state-of-the-art large language models such as GPT-4.

View Paper: https://www.semanticscholar.org/paper/cfce709a65f90312d2bdc1a6cf0380c19becf694

Citations in the corpus

• ReMindRAG: Low-Cost LLM-Guided Knowledge Graph Traversal for Efficient RAG (2025, unknown): “Retrieval-Augmented Generation (RAG) [23] has become a prominent method for augmenting Large Language Models (LLMs) with external knowledge resources, enabling them to generate more accurate outputs and thereby expanding their practical utility [17, 32].”
• Turk-LettuceDetect: A Hallucination Detection Model for Turkish RAG Applications (2025, unknown): “We utilized the RAGTruth dataset [6] for both training and evaluation of our models.”
• FreshStack: Building Realistic Benchmarks for Evaluating Retrieval on Technical Documents (2025, unknown): “Existing IR and RAG benchmarks are stale, evaluating on academic question answering datasets [20, 63, 61, 71, 86], or are not challenging, being constructed for RAG [87, 7, 40, 51, 71, 34].”