@inproceedings{jang-kim-2026-price,
title = "\texttt{{PRICE}}: Preemptive Confidence Estimation in {LLM}s via Internal Representations",
author = "Jang, Jonghak and
Kim, Misuk",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1445/",
pages = "32993--33018",
ISBN = "979-8-89176-459-0",
abstract = "Large language models have demonstrated strong performance across a wide range of tasks, yet they remain prone to generating incorrect information with high confidence. To mitigate this reliability issue, it is necessary to properly assess model confidence before answer generation. However, existing methods either rely on post-hoc estimation based on generated tokens or sampling results, or require additional training based on supervised learning, which limits their computational efficiency and generalizability. In this paper, we propose \textbf{PR}eemptive \textbf{I}nternal \textbf{C}onfidence \textbf{E}stimator (\texttt{PRICE}), which estimates confidence before answer generation by leveraging internal signals of LLMs that can be extracted from a single forward pass. \texttt{PRICE} combines signals that capture pattern agreement across attention heads and the semantic clarity of hidden states. Through experiments on diverse models and benchmarks, we show that \texttt{PRICE} achieves competitive performance against classifier-based methods without requiring additional training. These results suggest that \texttt{PRICE} can contribute to improving both the efficiency of LLMs and their safety in high-risk domains."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jang-kim-2026-price">
<titleInfo>
<title>PRICE: Preemptive Confidence Estimation in LLMs via Internal Representations</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jonghak</namePart>
<namePart type="family">Jang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Misuk</namePart>
<namePart type="family">Kim</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Large language models have demonstrated strong performance across a wide range of tasks, yet they remain prone to generating incorrect information with high confidence. To mitigate this reliability issue, it is necessary to properly assess model confidence before answer generation. However, existing methods either rely on post-hoc estimation based on generated tokens or sampling results, or require additional training based on supervised learning, which limits their computational efficiency and generalizability. In this paper, we propose PReemptive Internal Confidence Estimator (PRICE), which estimates confidence before answer generation by leveraging internal signals of LLMs that can be extracted from a single forward pass. PRICE combines signals that capture pattern agreement across attention heads and the semantic clarity of hidden states. Through experiments on diverse models and benchmarks, we show that PRICE achieves competitive performance against classifier-based methods without requiring additional training. These results suggest that PRICE can contribute to improving both the efficiency of LLMs and their safety in high-risk domains.</abstract>
<identifier type="citekey">jang-kim-2026-price</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1445/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>32993</start>
<end>33018</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T PRICE: Preemptive Confidence Estimation in LLMs via Internal Representations
%A Jang, Jonghak
%A Kim, Misuk
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F jang-kim-2026-price
%X Large language models have demonstrated strong performance across a wide range of tasks, yet they remain prone to generating incorrect information with high confidence. To mitigate this reliability issue, it is necessary to properly assess model confidence before answer generation. However, existing methods either rely on post-hoc estimation based on generated tokens or sampling results, or require additional training based on supervised learning, which limits their computational efficiency and generalizability. In this paper, we propose PReemptive Internal Confidence Estimator (PRICE), which estimates confidence before answer generation by leveraging internal signals of LLMs that can be extracted from a single forward pass. PRICE combines signals that capture pattern agreement across attention heads and the semantic clarity of hidden states. Through experiments on diverse models and benchmarks, we show that PRICE achieves competitive performance against classifier-based methods without requiring additional training. These results suggest that PRICE can contribute to improving both the efficiency of LLMs and their safety in high-risk domains.
%U https://aclanthology.org/2026.emnlp-main.1445/
%P 32993-33018
Markdown (Informal)
[PRICE: Preemptive Confidence Estimation in LLMs via Internal Representations](https://aclanthology.org/2026.emnlp-main.1445/) (Jang & Kim, EMNLP 2026)
ACL