@inproceedings{li-2026-geometry,
title = "The Geometry of Uncertainty in Language Model Representations",
author = "Li, Zy",
editor = "Zhang, Yue and
Xu, Ruifeng and
Purwarianti, Ayu and
Rong, Tong and
Suzuki, Jun",
booktitle = "Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the {A}sia-Pacific Chapter of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = nov,
year = "2026",
address = "Hengqin, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.aacl-long.122/",
pages = "2683--2702",
ISBN = "979-8-89176-439-2",
abstract = "Hidden-state probes have been used to predict factual errors in language-model outputs. We study whether the same representations also distinguish four types of prompt-model relationship: known, unknown, ambiguous, and misleading (built on a false premise). On a controlled 200-prompt dataset constructed to balance six surface features, all six pairwise type contrasts are linearly recoverable from hidden states at the position where generation begins, at AUROC $\geq 0.986$ on three open-weight instruction-tuned models spanning two families and a fourfold parameter range. The probe exceeds surface-feature, TF-IDF, and frozen sentence-embedding baselines on every pair. Under group-aware evaluation on paraphrased prompts with group-level shuffled controls, probe AUROC remains at least 0.949 on every pair while lexical baselines degrade. These findings frame hallucination as one downstream consequence of a four-type representational structure that also captures ambiguity and false premises."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="li-2026-geometry">
<titleInfo>
<title>The Geometry of Uncertainty in Language Model Representations</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruifeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ayu</namePart>
<namePart type="family">Purwarianti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Rong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Suzuki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hengqin, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-439-2</identifier>
</relatedItem>
<abstract>Hidden-state probes have been used to predict factual errors in language-model outputs. We study whether the same representations also distinguish four types of prompt-model relationship: known, unknown, ambiguous, and misleading (built on a false premise). On a controlled 200-prompt dataset constructed to balance six surface features, all six pairwise type contrasts are linearly recoverable from hidden states at the position where generation begins, at AUROC \geq 0.986 on three open-weight instruction-tuned models spanning two families and a fourfold parameter range. The probe exceeds surface-feature, TF-IDF, and frozen sentence-embedding baselines on every pair. Under group-aware evaluation on paraphrased prompts with group-level shuffled controls, probe AUROC remains at least 0.949 on every pair while lexical baselines degrade. These findings frame hallucination as one downstream consequence of a four-type representational structure that also captures ambiguity and false premises.</abstract>
<identifier type="citekey">li-2026-geometry</identifier>
<location>
<url>https://aclanthology.org/2026.aacl-long.122/</url>
</location>
<part>
<date>2026-11</date>
<extent unit="page">
<start>2683</start>
<end>2702</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T The Geometry of Uncertainty in Language Model Representations
%A Li, Zy
%Y Zhang, Yue
%Y Xu, Ruifeng
%Y Purwarianti, Ayu
%Y Rong, Tong
%Y Suzuki, Jun
%S Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 November
%I Association for Computational Linguistics
%C Hengqin, China
%@ 979-8-89176-439-2
%F li-2026-geometry
%X Hidden-state probes have been used to predict factual errors in language-model outputs. We study whether the same representations also distinguish four types of prompt-model relationship: known, unknown, ambiguous, and misleading (built on a false premise). On a controlled 200-prompt dataset constructed to balance six surface features, all six pairwise type contrasts are linearly recoverable from hidden states at the position where generation begins, at AUROC \geq 0.986 on three open-weight instruction-tuned models spanning two families and a fourfold parameter range. The probe exceeds surface-feature, TF-IDF, and frozen sentence-embedding baselines on every pair. Under group-aware evaluation on paraphrased prompts with group-level shuffled controls, probe AUROC remains at least 0.949 on every pair while lexical baselines degrade. These findings frame hallucination as one downstream consequence of a four-type representational structure that also captures ambiguity and false premises.
%U https://aclanthology.org/2026.aacl-long.122/
%P 2683-2702
Markdown (Informal)
[The Geometry of Uncertainty in Language Model Representations](https://aclanthology.org/2026.aacl-long.122/) (Li, AACL-IJCNLP 2026)
ACL
- Zy Li. 2026. The Geometry of Uncertainty in Language Model Representations. In Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 2683–2702, Hengqin, China. Association for Computational Linguistics.