analiza aiqbench
This research identifies relational linearity in embedding spaces as a strong predictor of hallucinations when language models encounter synthetic, unknown entities. By demonstrating that linear relations allow models to easily fabricate plausible but false facts for non-existent subjects, the study offers a new theoretical lens for understanding and mitigating this core failure mode in LLMs. The findings suggest that architectural shifts toward nonlinear representations could potentially reduce susceptibility to such hallucinations.
Relational Linearity is a Predictor of Hallucinations
arXiv:2601.11429v3 Announce Type: replace-cross Abstract: Hallucination is a central failure mode of language models (LMs). We focus on hallucinations in response to questions like: "Which instrument did Glenn Gould play?", but we ask these questions for synthetic entities designed to be unknown to the model. We find that LMs like Gemma-7B-IT frequently hallucinate, i.e., they have difficulty recognizing that the hallucinated fact is not part of their knowledge. Based on the idea of linear relational embeddings, we put forward the following hypothesis. (i) Due to the abstract scheme that is used to represent them, LMs can easily produce plausible objects for non-existing subjects of linear relations, which can lead to hallucinations. (ii) For nonlinear relations, this mechanism for producing an object is not available and so a hallucination is easier to avoid. To test this hypothesis, we create SynthHal, a synthetic unknown-entity benchmark for 15 relations. We find that across four instruction-tuned models, relational linearity is a strong predictor of models hallucinating an object for an unknown subject vs refusing to give an answer, with correlations $r \in [.58, .84]$. While this is not direct evidence for the hypothesized causal mechanism, it is suggestive and opens up a new line of inquiry into understanding LM hallucinations.
Abstract:Hallucination is a central failure mode of language models (LMs). We focus on hallucinations in response to questions like: "Which instrument did Glenn Gould play?", but we ask these questions for synthetic entities designed to be unknown to the model. We find that LMs like Gemma-7B-IT frequently hallucinate, i.e., they have difficulty recognizing that the hallucinated fact is not part of their knowledge. Based on the idea of linear relational embeddings, we put forward the following hypothesis. (i) Due to the abstract scheme that is used to represent them, LMs can easily produce plausible objects for non-existing subjects of linear relations, which can lead to hallucinations. (ii) For nonlinear relations, this mechanism for producing an object is not available and so a hallucination is easier to avoid. To test this hypothesis, we create SynthHal, a synthetic unknown-entity benchmark for 15 relations. We find that across four instruction-tuned models, relational linearity is a strong predictor of models hallucinating an object for an unknown subject vs refusing to give an answer, with correlations $r \in [.58, .84]$. While this is not direct evidence for the hypothesized causal mechanism, it is suggestive and opens up a new line of inquiry into understanding LM hallucinations.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
arXiv:2601.11429 [cs.CL]
(or arXiv:2601.11429v3 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2601.11429 Focus to learn more arXiv-issued DOI via DataCite
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 · 297 cuvinte
Subiecte conexe
Relational Linearity is a Predictor of Hallucinations
arXiv · 3 iun. 2026
Forecasting Bacterial Antimicrobial Resistance Trends Using Machine Learning on WHO GLASS Surveillance Data: A Retrieval-Augmented Generation Approach for Policy Decision Support
arXiv · 30 sept. 2026
The Entangled Dimensions of Decision Theory
LessWrong · 30 iul. 2026
PIRAMID: Progress and Plans
LessWrong · 27 iul. 2026