<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="leino-karlgren-2025-controlling">
    <titleInfo>
        <title>Controlling Language and Style of Multi-lingual Generative Language Models with Control Vectors</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Julius</namePart>
        <namePart type="family">Leino</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jussi</namePart>
        <namePart type="family">Karlgren</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-12</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <genre authority="bibutilsgt">journal article</genre>
    <relatedItem type="host">
        <titleInfo>
            <title>Northern European Journal of Language Technology</title>
        </titleInfo>
        <originInfo>
            <issuance>continuing</issuance>
            <publisher>Linköping University Electronic Press</publisher>
            <place>
                <placeTerm type="text">Linköping, Sweden</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">periodical</genre>
        <genre authority="bibutilsgt">academic journal</genre>
    </relatedItem>
    <abstract>Control vectors have recently gained popularity as a method for steering transformer-based generative language models. This paper contributes to this path of research by evaluating the robustness of these control vectors in multi- and cross-lingual question-answering settings mimicking the real-world deployment scenario, where models are expected to generate answers to challenging questions. We present a set of experiments to demonstrate that a control vector approach can be used to shift the output of a generative language model from one language to another, and to exercise stylistic control of the output across languages. Overall, we find that the control vector approach offers a relatively lightweight and effective path for developing methods to control the output of multilingual language models with multiple design choices affecting the real-world control performance.</abstract>
    <identifier type="citekey">leino-karlgren-2025-controlling</identifier>
    <identifier type="doi">10.3384/nejlt.2000-1533.2025.5888</identifier>
    <location>
        <url>https://aclanthology.org/2025.nejlt-1.1/</url>
    </location>
    <part>
        <date>2025-12</date>
        <detail type="volume"><number>11</number></detail>
        <extent unit="page">
            <start>1</start>
            <end>26</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Journal Article
%T Controlling Language and Style of Multi-lingual Generative Language Models with Control Vectors
%A Leino, Julius
%A Karlgren, Jussi
%J Northern European Journal of Language Technology
%D 2025
%8 December
%V 11
%I Linköping University Electronic Press
%C Linköping, Sweden
%F leino-karlgren-2025-controlling
%X Control vectors have recently gained popularity as a method for steering transformer-based generative language models. This paper contributes to this path of research by evaluating the robustness of these control vectors in multi- and cross-lingual question-answering settings mimicking the real-world deployment scenario, where models are expected to generate answers to challenging questions. We present a set of experiments to demonstrate that a control vector approach can be used to shift the output of a generative language model from one language to another, and to exercise stylistic control of the output across languages. Overall, we find that the control vector approach offers a relatively lightweight and effective path for developing methods to control the output of multilingual language models with multiple design choices affecting the real-world control performance.
%R 10.3384/nejlt.2000-1533.2025.5888
%U https://aclanthology.org/2025.nejlt-1.1/
%U https://doi.org/10.3384/nejlt.2000-1533.2025.5888
%P 1-26

Markdown (Informal)

[Controlling Language and Style of Multi-lingual Generative Language Models with Control Vectors](https://aclanthology.org/2025.nejlt-1.1/) (Leino & Karlgren, NEJLT 2025)

• Controlling Language and Style of Multi-lingual Generative Language Models with Control Vectors (Leino & Karlgren, NEJLT 2025)

ACL

• Julius Leino and Jussi Karlgren. 2025. Controlling Language and Style of Multi-lingual Generative Language Models with Control Vectors. Northern European Journal of Language Technology, 11:1–26.