Metraipolis › archive › dc_77fadbbc056f
Geometric Uncertainty for Detecting and Correcting Hallucinations in LLMs — arXiv:2509.13813
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2509.13813
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T10:57:44.746Z
- SHA-256 of the archived text
e4c0de61ae251aaeef361a36e2d9da4ce419b1a0427045c2011bdab30c824ae0- Stable URL of this page
https://metraipolis.com/documento/dc_77fadbbc056f
How to cite
Geometric Uncertainty for Detecting and Correcting Hallucinations in LLMs — arXiv:2509.13813 — arXiv (en), https://arxiv.org/abs/2509.13813 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_77fadbbc056f (Metraipolis archive dc_77fadbbc056f, sha256 e4c0de61ae251aaeef361a36e2d9da4ce419b1a0427045c2011bdab30c824ae0)
BibTeX
@misc{dc77fadbbc056f,
title = {Geometric Uncertainty for Detecting and Correcting Hallucinations in LLMs — arXiv:2509.13813},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2509.13813},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_77fadbbc056f; sha256 e4c0de61ae251aaeef361a36e2d9da4ce419b1a0427045c2011bdab30c824ae0},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2509.13813. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Large language models are known to hallucinate, generating linguistically plausible but incorrect answers to questions. Uncertainty quantification has been proposed as a strategy to detect such behaviour, but existing methods lack a unified framework to assess reliability at both the prompt and answer level. We introduce a geometric framework which quantifies language model uncertainty at both levels by explicitly modelling a prompt-conditioned semantic distribution in answer embedding space. Our approach is black-box and sampling-based; we generate multiple answers per prompt, and use archetypal analysis to estimate a geometric support for the answer distribution. At the prompt level, we approximate the distribution entropy to quantify uncertainty; for each individual answer, we then use notions of atypicality to assess its reliability relative to the batch. We employ our framework to not only detect hallucinations but correct them, by selecting the batch example deemed most reliable. Experiments show that our framework performs comparably to or better than prior methods on short form question-answering datasets, and achieves superior results on medical datasets where hallucinations carry particularly critical risks. Beyond pure performance, we suggest the theoretical grounding of our work provides support for semantic distributions as useful objects of study for language model uncertainty. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-24 in cs.LG). Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2509.13813 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-08T10:06:16.786Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).