Exécution de notre outil de synthèse sur un nouvel article
Vérifiez l’article ci-dessous, puis lancez sa synthèse.
HalluGuard: Evidence-Grounded Small Reasoning Models to Mitigate Hallucinations in Retrieval-Augmented Generation
Loris Bergeron (University of Luxembourg, Interdisciplinary Centre for Security, Reliability and Trust), Ioana Buhnila (Centre National de la Recherche Scientifique, Chosun University), Jérôme François (University of Luxembourg, Interdisciplinary Centre for Security, Reliability and Trust), Radu State (University of Luxembourg, Interdisciplinary Centre for Security, Reliability and Trust)
Résumé : Large Language Models excel at NLP tasks but remain prone to hallucinations, limiting trust in real-world applications.We present HalluGuard, a 4B-parameter Small Reasoning Model (SRM) designed as a guardrail for Retrieval-Augmented Generation (RAG) pipelines, which classifies document-claim pairs as grounded or hallucinated in closedbook, document-grounded settings and produces evidence-grounded justifications.Our approach combines (i) a domain-agnostic synthetic dataset derived from FineWeb and refined through multi-stage curation and data reformation, (ii) synthetic grounded and hallucinated claims, and (iii) preference-based finetuning with Odds Ratio Preference Optimization (ORPO) to distill large-model reasoning into a smaller backbone.On the RAGTruth subset of the LLM-AggreFact benchmark, HalluGuard achieves 84.4% balanced accuracy (BAcc), surpassing specialized models, MiniCheck (7B; 84.0%) and Granite Guardian 3.3 (8B; 82.2%) while using roughly half their parameters.Across the benchmark, it reaches 77.1% BAcc, surpassing larger general-purpose LLMs such as GPT-4o (75.9%).HalluGuard and its datasets are available on Hugging Face 1 .