@inproceedings{passinato-etal-2026-portuguese,
title = "{P}ortuguese-{IFE}val: Instruction-Following Evaluation in {P}ortuguese",
author = "Passinato, Emanuel Borges and
Fontana, Gustavo Pires and
Botelho, Cibelle Pires and
Biancolini, J{\'u}lia Jaime and
Silva, Diogo Fernandes Costa and
Galv{\~a}o Filho, Arlindo Rodrigues",
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.742/",
pages = "17210--17228",
ISBN = "979-8-89176-459-0",
abstract = "Instruction-following is a core capability of Large Language Models (LLMs), as they must reliably comply with instructions constraints before being able to be evaluated on downstream tasks. In this scenario, instruction-following benchmarks for low-resource languages remain limited, particularly for Portuguese, where existing evaluations are mostly based on translated English datasets and fail to capture language specific linguistic phenomena. We introduce Portuguese-IFEval, a benchmark for evaluating instruction following in Portuguese built through semantic regionalization. The benchmark contains a total of 1236 instructions across 535 prompts, each prompt have up to three verifiable instructions, combining adapted instructions with Portuguese-specific orthographic, grammatical, and discourse-level constraints. We evaluate 25 proprietary, open-source, and Portuguese-tuned LLMs under a unified protocol, establishing a comprehensive benchmark for Portuguese instruction-following evaluation. Our results show that, while English and Spanish remain the strongest-performing languages, Portuguese still presents substantial challenges, which are effectively exposed by our benchmark. Our code is available at: https://github.com/gap-ufg/Portuguese-IFEval."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="passinato-etal-2026-portuguese">
<titleInfo>
<title>Portuguese-IFEval: Instruction-Following Evaluation in Portuguese</title>
</titleInfo>
<name type="personal">
<namePart type="given">Emanuel</namePart>
<namePart type="given">Borges</namePart>
<namePart type="family">Passinato</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gustavo</namePart>
<namePart type="given">Pires</namePart>
<namePart type="family">Fontana</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cibelle</namePart>
<namePart type="given">Pires</namePart>
<namePart type="family">Botelho</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Júlia</namePart>
<namePart type="given">Jaime</namePart>
<namePart type="family">Biancolini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Diogo</namePart>
<namePart type="given">Fernandes</namePart>
<namePart type="given">Costa</namePart>
<namePart type="family">Silva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Arlindo</namePart>
<namePart type="given">Rodrigues</namePart>
<namePart type="family">Galvão Filho</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>Instruction-following is a core capability of Large Language Models (LLMs), as they must reliably comply with instructions constraints before being able to be evaluated on downstream tasks. In this scenario, instruction-following benchmarks for low-resource languages remain limited, particularly for Portuguese, where existing evaluations are mostly based on translated English datasets and fail to capture language specific linguistic phenomena. We introduce Portuguese-IFEval, a benchmark for evaluating instruction following in Portuguese built through semantic regionalization. The benchmark contains a total of 1236 instructions across 535 prompts, each prompt have up to three verifiable instructions, combining adapted instructions with Portuguese-specific orthographic, grammatical, and discourse-level constraints. We evaluate 25 proprietary, open-source, and Portuguese-tuned LLMs under a unified protocol, establishing a comprehensive benchmark for Portuguese instruction-following evaluation. Our results show that, while English and Spanish remain the strongest-performing languages, Portuguese still presents substantial challenges, which are effectively exposed by our benchmark. Our code is available at: https://github.com/gap-ufg/Portuguese-IFEval.</abstract>
<identifier type="citekey">passinato-etal-2026-portuguese</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.742/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>17210</start>
<end>17228</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Portuguese-IFEval: Instruction-Following Evaluation in Portuguese
%A Passinato, Emanuel Borges
%A Fontana, Gustavo Pires
%A Botelho, Cibelle Pires
%A Biancolini, Júlia Jaime
%A Silva, Diogo Fernandes Costa
%A Galvão Filho, Arlindo Rodrigues
%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 passinato-etal-2026-portuguese
%X Instruction-following is a core capability of Large Language Models (LLMs), as they must reliably comply with instructions constraints before being able to be evaluated on downstream tasks. In this scenario, instruction-following benchmarks for low-resource languages remain limited, particularly for Portuguese, where existing evaluations are mostly based on translated English datasets and fail to capture language specific linguistic phenomena. We introduce Portuguese-IFEval, a benchmark for evaluating instruction following in Portuguese built through semantic regionalization. The benchmark contains a total of 1236 instructions across 535 prompts, each prompt have up to three verifiable instructions, combining adapted instructions with Portuguese-specific orthographic, grammatical, and discourse-level constraints. We evaluate 25 proprietary, open-source, and Portuguese-tuned LLMs under a unified protocol, establishing a comprehensive benchmark for Portuguese instruction-following evaluation. Our results show that, while English and Spanish remain the strongest-performing languages, Portuguese still presents substantial challenges, which are effectively exposed by our benchmark. Our code is available at: https://github.com/gap-ufg/Portuguese-IFEval.
%U https://aclanthology.org/2026.emnlp-main.742/
%P 17210-17228
Markdown (Informal)
[Portuguese-IFEval: Instruction-Following Evaluation in Portuguese](https://aclanthology.org/2026.emnlp-main.742/) (Passinato et al., EMNLP 2026)
ACL
- Emanuel Borges Passinato, Gustavo Pires Fontana, Cibelle Pires Botelho, Júlia Jaime Biancolini, Diogo Fernandes Costa Silva, and Arlindo Rodrigues Galvão Filho. 2026. Portuguese-IFEval: Instruction-Following Evaluation in Portuguese. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 17210–17228, Budapest, Hungary. Association for Computational Linguistics.