@inproceedings{shi-etal-2026-sage-answer,
title = "{SAGE}: Answer-Conditioned Uncertainty Targets for Verbal Confidence Alignment",
author = "Shi, Kaiwen and
Zhang, Zheyuan and
Ye, Yanfang",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1214/",
pages = "25717--25737",
ISBN = "979-8-89176-462-0",
abstract = "Large language models increasingly express uncertainty through verbalized confidence, yet these expressions often fail to reflect the model{'}s sampled behavior. We study verbal confidence alignment as a distributional calibration problem, where the confidence target for a prompt is estimated from repeated model outputs rather than from a single response. However, group rollouts alone are insufficient, since the resulting target must provide a useful training signal. We propose Semantic-Answer Guided Entropy (SAGE), a group-level uncertainty target that constructs an answer-conditioned uncertainty geometry over sampled responses. SAGE preserves categorical, numeric, and symbolic answer distinctions while maintaining a smooth and scale-preserving calibration signal. We further introduce Group-Uncertainty Preference Optimization (GUPO), an uncertainty-channel training framework that supervises verbal confidence expressions rather than the full response. Experiments across factual, mathematical, and multiple-choice reasoning tasks show improved confidence ranking, lower calibration error, and reduced overconfidence."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shi-etal-2026-sage-answer">
<titleInfo>
<title>SAGE: Answer-Conditioned Uncertainty Targets for Verbal Confidence Alignment</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kaiwen</namePart>
<namePart type="family">Shi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zheyuan</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanfang</namePart>
<namePart type="family">Ye</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Large language models increasingly express uncertainty through verbalized confidence, yet these expressions often fail to reflect the model’s sampled behavior. We study verbal confidence alignment as a distributional calibration problem, where the confidence target for a prompt is estimated from repeated model outputs rather than from a single response. However, group rollouts alone are insufficient, since the resulting target must provide a useful training signal. We propose Semantic-Answer Guided Entropy (SAGE), a group-level uncertainty target that constructs an answer-conditioned uncertainty geometry over sampled responses. SAGE preserves categorical, numeric, and symbolic answer distinctions while maintaining a smooth and scale-preserving calibration signal. We further introduce Group-Uncertainty Preference Optimization (GUPO), an uncertainty-channel training framework that supervises verbal confidence expressions rather than the full response. Experiments across factual, mathematical, and multiple-choice reasoning tasks show improved confidence ranking, lower calibration error, and reduced overconfidence.</abstract>
<identifier type="citekey">shi-etal-2026-sage-answer</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1214/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>25717</start>
<end>25737</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SAGE: Answer-Conditioned Uncertainty Targets for Verbal Confidence Alignment
%A Shi, Kaiwen
%A Zhang, Zheyuan
%A Ye, Yanfang
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F shi-etal-2026-sage-answer
%X Large language models increasingly express uncertainty through verbalized confidence, yet these expressions often fail to reflect the model’s sampled behavior. We study verbal confidence alignment as a distributional calibration problem, where the confidence target for a prompt is estimated from repeated model outputs rather than from a single response. However, group rollouts alone are insufficient, since the resulting target must provide a useful training signal. We propose Semantic-Answer Guided Entropy (SAGE), a group-level uncertainty target that constructs an answer-conditioned uncertainty geometry over sampled responses. SAGE preserves categorical, numeric, and symbolic answer distinctions while maintaining a smooth and scale-preserving calibration signal. We further introduce Group-Uncertainty Preference Optimization (GUPO), an uncertainty-channel training framework that supervises verbal confidence expressions rather than the full response. Experiments across factual, mathematical, and multiple-choice reasoning tasks show improved confidence ranking, lower calibration error, and reduced overconfidence.
%U https://aclanthology.org/2026.findings-emnlp.1214/
%P 25717-25737
Markdown (Informal)
[SAGE: Answer-Conditioned Uncertainty Targets for Verbal Confidence Alignment](https://aclanthology.org/2026.findings-emnlp.1214/) (Shi et al., Findings 2026)
ACL