@inproceedings{shore-etal-2026-portnlp,
title = "{P}ort{NLP} at {CRAC} 2026: {QL}o{RA} Fine-Tuning with Bounded Entity Registry for Multilingual Coreference Resolution",
author = "Shore, Amber and
Scheinberg, Russell and
Nagasundaram, Malini and
Agrawal, Ameeta",
editor = "Braud, Chlo{\'e} and
Hardmeier, Christian and
Ogrodniczuk, Maciej and
Loaiciga, Sharid and
Zeldes, Amir and
Nov{\'a}k, Michal and
Li, Chuyuan and
Strube, Michael and
Li, Junyi Jessy",
booktitle = "Proceedings of the 2nd Joint Workshop on Computational Approaches to Discourse, Context and Document-Level Inferences and Computational Models of Reference, Anaphora and Coreference ({CODI}-{CRAC} 2026)",
month = jul,
year = "2026",
address = "San Diego, California, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.codi-1.25/",
pages = "193--198",
ISBN = "979-8-89176-400-2",
abstract = "We describe PortNLP{'}s submission to the CRAC 2026 Shared Task on Multilingual Coreference Resolution (LLM track). Our system fine-tunes Qwen 3 14B with QLoRA on CorefUD 1.4 gold annotations across 27 corpora spanning 19 languages. Documents are processed in 500-700 character chunks with a bounded rolling context consisting of 500 characters of recent annotated text and a scored entity registry that tracks up to 30 active entities via a frequency-times-recency decay formula. We employ data augmentation and language-aware sampling strategies to handle typological and data-size diversity. Our system achieves 68.69 CoNLL F1 averaged across all 27 test corpora. We additionally present probing experiments on the LoRA adapter{'}s internal representations, finding that coreference signal is concentrated in attention value projections rather than MLP modules, with the strongest readout at the earliest transformer layer."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shore-etal-2026-portnlp">
<titleInfo>
<title>PortNLP at CRAC 2026: QLoRA Fine-Tuning with Bounded Entity Registry for Multilingual Coreference Resolution</title>
</titleInfo>
<name type="personal">
<namePart type="given">Amber</namePart>
<namePart type="family">Shore</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Russell</namePart>
<namePart type="family">Scheinberg</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Malini</namePart>
<namePart type="family">Nagasundaram</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ameeta</namePart>
<namePart type="family">Agrawal</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 2nd Joint Workshop on Computational Approaches to Discourse, Context and Document-Level Inferences and Computational Models of Reference, Anaphora and Coreference (CODI-CRAC 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chloé</namePart>
<namePart type="family">Braud</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christian</namePart>
<namePart type="family">Hardmeier</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maciej</namePart>
<namePart type="family">Ogrodniczuk</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sharid</namePart>
<namePart type="family">Loaiciga</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Amir</namePart>
<namePart type="family">Zeldes</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Michal</namePart>
<namePart type="family">Novák</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chuyuan</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Michael</namePart>
<namePart type="family">Strube</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junyi</namePart>
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</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, USA</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-400-2</identifier>
</relatedItem>
<abstract>We describe PortNLP’s submission to the CRAC 2026 Shared Task on Multilingual Coreference Resolution (LLM track). Our system fine-tunes Qwen 3 14B with QLoRA on CorefUD 1.4 gold annotations across 27 corpora spanning 19 languages. Documents are processed in 500-700 character chunks with a bounded rolling context consisting of 500 characters of recent annotated text and a scored entity registry that tracks up to 30 active entities via a frequency-times-recency decay formula. We employ data augmentation and language-aware sampling strategies to handle typological and data-size diversity. Our system achieves 68.69 CoNLL F1 averaged across all 27 test corpora. We additionally present probing experiments on the LoRA adapter’s internal representations, finding that coreference signal is concentrated in attention value projections rather than MLP modules, with the strongest readout at the earliest transformer layer.</abstract>
<identifier type="citekey">shore-etal-2026-portnlp</identifier>
<location>
<url>https://aclanthology.org/2026.codi-1.25/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>193</start>
<end>198</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T PortNLP at CRAC 2026: QLoRA Fine-Tuning with Bounded Entity Registry for Multilingual Coreference Resolution
%A Shore, Amber
%A Scheinberg, Russell
%A Nagasundaram, Malini
%A Agrawal, Ameeta
%Y Braud, Chloé
%Y Hardmeier, Christian
%Y Ogrodniczuk, Maciej
%Y Loaiciga, Sharid
%Y Zeldes, Amir
%Y Novák, Michal
%Y Li, Chuyuan
%Y Strube, Michael
%Y Li, Junyi Jessy
%S Proceedings of the 2nd Joint Workshop on Computational Approaches to Discourse, Context and Document-Level Inferences and Computational Models of Reference, Anaphora and Coreference (CODI-CRAC 2026)
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, USA
%@ 979-8-89176-400-2
%F shore-etal-2026-portnlp
%X We describe PortNLP’s submission to the CRAC 2026 Shared Task on Multilingual Coreference Resolution (LLM track). Our system fine-tunes Qwen 3 14B with QLoRA on CorefUD 1.4 gold annotations across 27 corpora spanning 19 languages. Documents are processed in 500-700 character chunks with a bounded rolling context consisting of 500 characters of recent annotated text and a scored entity registry that tracks up to 30 active entities via a frequency-times-recency decay formula. We employ data augmentation and language-aware sampling strategies to handle typological and data-size diversity. Our system achieves 68.69 CoNLL F1 averaged across all 27 test corpora. We additionally present probing experiments on the LoRA adapter’s internal representations, finding that coreference signal is concentrated in attention value projections rather than MLP modules, with the strongest readout at the earliest transformer layer.
%U https://aclanthology.org/2026.codi-1.25/
%P 193-198
Markdown (Informal)
[PortNLP at CRAC 2026: QLoRA Fine-Tuning with Bounded Entity Registry for Multilingual Coreference Resolution](https://aclanthology.org/2026.codi-1.25/) (Shore et al., CODI-CRAC 2026)
ACL