analiza aiqbench
The paper addresses cross-model hallucination detection: training a detector on one LLM's hidden states and applying it to unseen models. It proposes modelling hidden states as relational graphs over layers, tokens, and features, combined with meta-learning. The claimed gains rest on the authors' own experiments, and the code is linked but not yet independently validated. Readers building LLM reliability tooling should watch whether relational signals transfer beyond the evaluated model families.
CHORD: Cross-Model Hallucination Detection via Relational Graph Discrimination
arXiv:2610.38314v1 Announce Type: new Abstract: Hallucination detection is critical for deploying large language models (LLMs) in real-world applications. Due to strong empirical performance, internal representation-based methods have emerged as the prevailing direction for detecting hallucinations, yet they remain largely model-specific and often fail to generalize to unseen LLMs. In this paper, we study an important yet underexplored problem, termed cross-model hallucination detection (CMHD), which aims to train hallucination detectors on source LLMs while ensuring performance on unseen target LLMs. The core challenge of CMHD lies in cross-model representation heterogeneity: hidden states from different LLMs exhibit severe semantic inconsistency, which limits the transferability of classical representation-based detectors. Through analysis and empirical validation, we show that relational structures over layers, tokens, and features capture transferable detection signals despite semantic inconsistency. Based on this observation, we propose cross-model hallucination detection via heterogeneity-oriented relational discrimination (CHORD), a relational graph discrimination framework that represents hidden states as joint relational graphs over layers, tokens, and features. It feeds these graphs into relation-aware attention to obtain structure-aware features, and uses meta-learning to shape these features toward transferable detection signals. Extensive experiments show that CHORD outperforms representative detection methods in cross-model generalization. Code is available at https://github.com/MiracleworkerD/CHORD.
Abstract:Hallucination detection is critical for deploying large language models (LLMs) in real-world applications. Due to strong empirical performance, internal representation-based methods have emerged as the prevailing direction for detecting hallucinations, yet they remain largely model-specific and often fail to generalize to unseen LLMs. In this paper, we study an important yet underexplored problem, termed cross-model hallucination detection (CMHD), which aims to train hallucination detectors on source LLMs while ensuring performance on unseen target LLMs. The core challenge of CMHD lies in cross-model representation heterogeneity: hidden states from different LLMs exhibit severe semantic inconsistency, which limits the transferability of classical representation-based detectors. Through analysis and empirical validation, we show that relational structures over layers, tokens, and features capture transferable detection signals despite semantic inconsistency. Based on this observation, we propose cross-model hallucination detection via heterogeneity-oriented relational discrimination (CHORD), a relational graph discrimination framework that represents hidden states as joint relational graphs over layers, tokens, and features. It feeds these graphs into relation-aware attention to obtain structure-aware features, and uses meta-learning to shape these features toward transferable detection signals. Extensive experiments show that CHORD outperforms representative detection methods in cross-model generalization. Code is available at this https URL.
Accepted to NeurIPS 2026 as a Spotlight paper. 64 pages, 14 figures, 7 tables
Artificial Intelligence (cs.AI)
arXiv:2610.38314 [cs.AI]
(or arXiv:2610.38314v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.38314 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 304 cuvinte
Subiecte conexe
BenHalluEval: A Multi-Task Hallucination Evaluation Framework for Large Language Models on Bengali
arXiv · 4 iun. 2026
BenHalluEval: A Multi-Task Hallucination Evaluation Framework for Large Language Models on Bengali
arXiv · 7 aug. 2026
Mitigating Object Hallucinations in LVLMs via Attention Imbalance Rectification
arXiv · 16 iun. 2026
UniFGVC: Universal Training-Free Few-Shot Fine-Grained Vision Classification via Attribute-Aware Multimodal Retrieval
arXiv · 17 oct. 2026