<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="baek-etal-2025-revisiting">
    <titleInfo>
        <title>Revisiting In-Context Learning with Long Context Language Models</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Jinheon</namePart>
        <namePart type="family">Baek</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Sun</namePart>
        <namePart type="given">Jae</namePart>
        <namePart type="family">Lee</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Prakhar</namePart>
        <namePart type="family">Gupta</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Geunseob</namePart>
        <namePart type="family">Oh</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Siddharth</namePart>
        <namePart type="family">Dalmia</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Prateek</namePart>
        <namePart type="family">Kolhar</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: ACL 2025</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Wanxiang</namePart>
            <namePart type="family">Che</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Joyce</namePart>
            <namePart type="family">Nabende</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Ekaterina</namePart>
            <namePart type="family">Shutova</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Mohammad</namePart>
            <namePart type="given">Taher</namePart>
            <namePart type="family">Pilehvar</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Vienna, Austria</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-256-5</identifier>
    </relatedItem>
    <abstract>In-Context Learning (ICL) is a technique by which language models make predictions based on examples provided in their input context. Previously, their context window size imposed a limit on the number of examples that can be shown, making example selection techniques crucial for identifying the maximally effective set of examples. However, the recent advent of Long Context Language Models (LCLMs) has significantly increased the number of examples that can be included in context, raising an important question of whether ICL performance in a many-shot regime is still sensitive to the method of sample selection. To answer this, we revisit these approaches in the context of LCLMs through extensive experiments on 18 datasets spanning 4 tasks. Surprisingly, we observe that sophisticated example selection techniques do not yield significant improvements over a simple random sample selection method. Instead, we discover that the advent of LCLMs has fundamentally shifted the challenge of ICL from that of selecting the most effective examples to that of collecting sufficient examples to fill the context window. Specifically, in certain datasets, including all available examples does not fully utilize the context window; however, by augmenting the examples in context with a simple data augmentation approach, we substantially improve ICL performance by 5%.</abstract>
    <identifier type="citekey">baek-etal-2025-revisiting</identifier>
    <identifier type="doi">10.18653/v1/2025.findings-acl.1382</identifier>
    <location>
        <url>https://aclanthology.org/2025.findings-acl.1382/</url>
    </location>
    <part>
        <date>2025-07</date>
        <extent unit="page">
            <start>26950</start>
            <end>26966</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Revisiting In-Context Learning with Long Context Language Models
%A Baek, Jinheon
%A Lee, Sun Jae
%A Gupta, Prakhar
%A Oh, Geunseob
%A Dalmia, Siddharth
%A Kolhar, Prateek
%Y Che, Wanxiang
%Y Nabende, Joyce
%Y Shutova, Ekaterina
%Y Pilehvar, Mohammad Taher
%S Findings of the Association for Computational Linguistics: ACL 2025
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-256-5
%F baek-etal-2025-revisiting
%X In-Context Learning (ICL) is a technique by which language models make predictions based on examples provided in their input context. Previously, their context window size imposed a limit on the number of examples that can be shown, making example selection techniques crucial for identifying the maximally effective set of examples. However, the recent advent of Long Context Language Models (LCLMs) has significantly increased the number of examples that can be included in context, raising an important question of whether ICL performance in a many-shot regime is still sensitive to the method of sample selection. To answer this, we revisit these approaches in the context of LCLMs through extensive experiments on 18 datasets spanning 4 tasks. Surprisingly, we observe that sophisticated example selection techniques do not yield significant improvements over a simple random sample selection method. Instead, we discover that the advent of LCLMs has fundamentally shifted the challenge of ICL from that of selecting the most effective examples to that of collecting sufficient examples to fill the context window. Specifically, in certain datasets, including all available examples does not fully utilize the context window; however, by augmenting the examples in context with a simple data augmentation approach, we substantially improve ICL performance by 5%.
%R 10.18653/v1/2025.findings-acl.1382
%U https://aclanthology.org/2025.findings-acl.1382/
%U https://doi.org/10.18653/v1/2025.findings-acl.1382
%P 26950-26966

Markdown (Informal)

[Revisiting In-Context Learning with Long Context Language Models](https://aclanthology.org/2025.findings-acl.1382/) (Baek et al., Findings 2025)

• Revisiting In-Context Learning with Long Context Language Models (Baek et al., Findings 2025)

ACL

• Jinheon Baek, Sun Jae Lee, Prakhar Gupta, Geunseob Oh, Siddharth Dalmia, and Prateek Kolhar. 2025. Revisiting In-Context Learning with Long Context Language Models. In Findings of the Association for Computational Linguistics: ACL 2025, pages 26950–26966, Vienna, Austria. Association for Computational Linguistics.