@inproceedings{fonseca-etal-2026-high,
title = "When High Accuracy Hides Poor Calibration: Rethinking Confidence Evaluation in Transformer-Based Text Classification with Balanced Brier Score",
author = "Fonseca, Guilherme and
Prenassi, Gabriel and
Cunha, Washington and
Rocha, Leonardo Chaves Dutra da and
Gon{\c{c}}alves, Marcos Andr{\'e}",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.acl-long.2128/",
pages = "45888--45900",
ISBN = "979-8-89176-390-6",
abstract = "Transformer-based Small (SLMs) and Large Language Models (LLMs) achieve strong effectiveness in text classification (TC), yet deployment requires reliable confidence estimates. Although miscalibration in Transformers has been reported, evidence for TC under fine-tuning remains limited. We evaluate the calibration of fine-tuned SLMs and LLMs against Logistic Regression, a classical, well-calibrated baseline, and find that, despite superior effectiveness, Transformers remain markedly overconfident. Crucially, we show that widely used calibration metrics, such as Expected Calibration Error and Brier Score, become biased in high-effectiveness regimes, where the dominance of correct predictions masks severe miscalibration on errors, sometimes even suggesting better calibration than Logistic Regression, a well-known calibrated method. To address this limitation, we propose the Balanced Brier Score (BBS), which balances the contribution of correct and incorrect predictions within confidence bins. BBS reveals substantially poorer calibration in both SLMs and LLMs, consistent with qualitative evidence from calibration curves. These findings challenge current calibration assessment practices and provide a more reliable alternative for evaluating confidence quality in Transformer-based TC."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="fonseca-etal-2026-high">
<titleInfo>
<title>When High Accuracy Hides Poor Calibration: Rethinking Confidence Evaluation in Transformer-Based Text Classification with Balanced Brier Score</title>
</titleInfo>
<name type="personal">
<namePart type="given">Guilherme</namePart>
<namePart type="family">Fonseca</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gabriel</namePart>
<namePart type="family">Prenassi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Washington</namePart>
<namePart type="family">Cunha</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Leonardo</namePart>
<namePart type="given">Chaves</namePart>
<namePart type="given">Dutra</namePart>
<namePart type="given">da</namePart>
<namePart type="family">Rocha</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marcos</namePart>
<namePart type="given">André</namePart>
<namePart type="family">Gonçalves</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-390-6</identifier>
</relatedItem>
<abstract>Transformer-based Small (SLMs) and Large Language Models (LLMs) achieve strong effectiveness in text classification (TC), yet deployment requires reliable confidence estimates. Although miscalibration in Transformers has been reported, evidence for TC under fine-tuning remains limited. We evaluate the calibration of fine-tuned SLMs and LLMs against Logistic Regression, a classical, well-calibrated baseline, and find that, despite superior effectiveness, Transformers remain markedly overconfident. Crucially, we show that widely used calibration metrics, such as Expected Calibration Error and Brier Score, become biased in high-effectiveness regimes, where the dominance of correct predictions masks severe miscalibration on errors, sometimes even suggesting better calibration than Logistic Regression, a well-known calibrated method. To address this limitation, we propose the Balanced Brier Score (BBS), which balances the contribution of correct and incorrect predictions within confidence bins. BBS reveals substantially poorer calibration in both SLMs and LLMs, consistent with qualitative evidence from calibration curves. These findings challenge current calibration assessment practices and provide a more reliable alternative for evaluating confidence quality in Transformer-based TC.</abstract>
<identifier type="citekey">fonseca-etal-2026-high</identifier>
<location>
<url>https://aclanthology.org/2026.acl-long.2128/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>45888</start>
<end>45900</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T When High Accuracy Hides Poor Calibration: Rethinking Confidence Evaluation in Transformer-Based Text Classification with Balanced Brier Score
%A Fonseca, Guilherme
%A Prenassi, Gabriel
%A Cunha, Washington
%A Rocha, Leonardo Chaves Dutra da
%A Gonçalves, Marcos André
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-390-6
%F fonseca-etal-2026-high
%X Transformer-based Small (SLMs) and Large Language Models (LLMs) achieve strong effectiveness in text classification (TC), yet deployment requires reliable confidence estimates. Although miscalibration in Transformers has been reported, evidence for TC under fine-tuning remains limited. We evaluate the calibration of fine-tuned SLMs and LLMs against Logistic Regression, a classical, well-calibrated baseline, and find that, despite superior effectiveness, Transformers remain markedly overconfident. Crucially, we show that widely used calibration metrics, such as Expected Calibration Error and Brier Score, become biased in high-effectiveness regimes, where the dominance of correct predictions masks severe miscalibration on errors, sometimes even suggesting better calibration than Logistic Regression, a well-known calibrated method. To address this limitation, we propose the Balanced Brier Score (BBS), which balances the contribution of correct and incorrect predictions within confidence bins. BBS reveals substantially poorer calibration in both SLMs and LLMs, consistent with qualitative evidence from calibration curves. These findings challenge current calibration assessment practices and provide a more reliable alternative for evaluating confidence quality in Transformer-based TC.
%U https://aclanthology.org/2026.acl-long.2128/
%P 45888-45900
Markdown (Informal)
[When High Accuracy Hides Poor Calibration: Rethinking Confidence Evaluation in Transformer-Based Text Classification with Balanced Brier Score](https://aclanthology.org/2026.acl-long.2128/) (Fonseca et al., ACL 2026)
ACL