@inproceedings{kang-etal-2026-beyond-paraphrases,
title = "Beyond Paraphrases: Generalization Across Task Formats in Parametric Knowledge Editing",
author = "Kang, Cheongwoong and
Baek, Jongeun and
Kim, Yeonjea and
Choi, Jaesik",
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.1188/",
pages = "25089--25110",
ISBN = "979-8-89176-462-0",
abstract = "Knowledge editing aims to update specific facts in language models to ensure reliable and up-to-date behavior. A key challenge is whether edited knowledge generalizes beyond the format in which the edit is applied, yet existing evaluations primarily assess generalization using paraphrases. In this work, we identify a critical blind spot: generalization across heterogeneous task formats, including multiple-choice and verification-based question answering. We show that paraphrase-based evaluations overestimate robustness, as models that perform well on paraphrases exhibit substantial performance drops across task formats. While model predictions are consistent across formats prior to editing, editing introduces format-dependent inconsistencies. We further observe that heterogeneous task formats exhibit distinct decision landscapes, where single-format objectives lead to format-specific local optima with divergent edit directions, resulting in misaligned outputs across formats. To address this issue, we demonstrate that aligning edit directions across task formats improves cross-format generalization and consistency. These findings highlight the need to move beyond paraphrase-based evaluation and to develop editing methods that maintain consistent factual behavior across diverse task formulations."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kang-etal-2026-beyond-paraphrases">
<titleInfo>
<title>Beyond Paraphrases: Generalization Across Task Formats in Parametric Knowledge Editing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Cheongwoong</namePart>
<namePart type="family">Kang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jongeun</namePart>
<namePart type="family">Baek</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yeonjea</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jaesik</namePart>
<namePart type="family">Choi</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>Knowledge editing aims to update specific facts in language models to ensure reliable and up-to-date behavior. A key challenge is whether edited knowledge generalizes beyond the format in which the edit is applied, yet existing evaluations primarily assess generalization using paraphrases. In this work, we identify a critical blind spot: generalization across heterogeneous task formats, including multiple-choice and verification-based question answering. We show that paraphrase-based evaluations overestimate robustness, as models that perform well on paraphrases exhibit substantial performance drops across task formats. While model predictions are consistent across formats prior to editing, editing introduces format-dependent inconsistencies. We further observe that heterogeneous task formats exhibit distinct decision landscapes, where single-format objectives lead to format-specific local optima with divergent edit directions, resulting in misaligned outputs across formats. To address this issue, we demonstrate that aligning edit directions across task formats improves cross-format generalization and consistency. These findings highlight the need to move beyond paraphrase-based evaluation and to develop editing methods that maintain consistent factual behavior across diverse task formulations.</abstract>
<identifier type="citekey">kang-etal-2026-beyond-paraphrases</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1188/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>25089</start>
<end>25110</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond Paraphrases: Generalization Across Task Formats in Parametric Knowledge Editing
%A Kang, Cheongwoong
%A Baek, Jongeun
%A Kim, Yeonjea
%A Choi, Jaesik
%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 kang-etal-2026-beyond-paraphrases
%X Knowledge editing aims to update specific facts in language models to ensure reliable and up-to-date behavior. A key challenge is whether edited knowledge generalizes beyond the format in which the edit is applied, yet existing evaluations primarily assess generalization using paraphrases. In this work, we identify a critical blind spot: generalization across heterogeneous task formats, including multiple-choice and verification-based question answering. We show that paraphrase-based evaluations overestimate robustness, as models that perform well on paraphrases exhibit substantial performance drops across task formats. While model predictions are consistent across formats prior to editing, editing introduces format-dependent inconsistencies. We further observe that heterogeneous task formats exhibit distinct decision landscapes, where single-format objectives lead to format-specific local optima with divergent edit directions, resulting in misaligned outputs across formats. To address this issue, we demonstrate that aligning edit directions across task formats improves cross-format generalization and consistency. These findings highlight the need to move beyond paraphrase-based evaluation and to develop editing methods that maintain consistent factual behavior across diverse task formulations.
%U https://aclanthology.org/2026.findings-emnlp.1188/
%P 25089-25110
Markdown (Informal)
[Beyond Paraphrases: Generalization Across Task Formats in Parametric Knowledge Editing](https://aclanthology.org/2026.findings-emnlp.1188/) (Kang et al., Findings 2026)
ACL