<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jiao-etal-2025-analyzing">
    <titleInfo>
        <title>Analyzing and Modeling LLM Response Lengths with Extreme Value Theory: Anchoring Effects and Hybrid Distributions</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Liuxuan</namePart>
        <namePart type="family">Jiao</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Chen</namePart>
        <namePart type="family">Gao</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yiqian</namePart>
        <namePart type="family">Yang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Chenliang</namePart>
        <namePart type="family">Zhou</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">YiXian</namePart>
        <namePart type="family">Huang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Xinlei</namePart>
        <namePart type="family">Chen</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yong</namePart>
        <namePart type="family">Li</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-11</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Christos</namePart>
            <namePart type="family">Christodoulopoulos</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Tanmoy</namePart>
            <namePart type="family">Chakraborty</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Carolyn</namePart>
            <namePart type="family">Rose</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Violet</namePart>
            <namePart type="family">Peng</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Suzhou, China</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-332-6</identifier>
    </relatedItem>
    <abstract>We present a statistical framework for modeling and controlling large language model (LLM) response lengths using extreme value theory. Analyzing 14,301 GPT-4o responses across temperature and prompting conditions, with cross-validation on Qwen and DeepSeek architectures, we demonstrate that verbosity follows Weibull-type generalized extreme value (GEV) distributions with heavier tails under stochastic generation. Our key contributions include: (1) development of a novel GEV-generalized Pareto (GPD) hybrid model that improves tail fit (R²_\textCDF=0.9993 vs standalone GEV’s 0.998) while maintaining architectural generalizability; (2) quantitative characterization of prompt anchoring effects across models, showing reduced dispersion but increased outliers under randomization; and (3) identification of temperature-dependent response patterns that persist across architectures, with higher temperatures amplifying length variability while preserving extreme-value mechanisms. The hybrid model’s threshold selection method enables precise verbosity control in production systems regardless of model choice. While validated on multiple architectures, generalizability to emerging model families requires further study.</abstract>
    <identifier type="citekey">jiao-etal-2025-analyzing</identifier>
    <identifier type="doi">10.18653/v1/2025.emnlp-main.1676</identifier>
    <location>
        <url>https://aclanthology.org/2025.emnlp-main.1676/</url>
    </location>
    <part>
        <date>2025-11</date>
        <extent unit="page">
            <start>32992</start>
            <end>33002</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Analyzing and Modeling LLM Response Lengths with Extreme Value Theory: Anchoring Effects and Hybrid Distributions
%A Jiao, Liuxuan
%A Gao, Chen
%A Yang, Yiqian
%A Zhou, Chenliang
%A Huang, YiXian
%A Chen, Xinlei
%A Li, Yong
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-332-6
%F jiao-etal-2025-analyzing
%X We present a statistical framework for modeling and controlling large language model (LLM) response lengths using extreme value theory. Analyzing 14,301 GPT-4o responses across temperature and prompting conditions, with cross-validation on Qwen and DeepSeek architectures, we demonstrate that verbosity follows Weibull-type generalized extreme value (GEV) distributions with heavier tails under stochastic generation. Our key contributions include: (1) development of a novel GEV-generalized Pareto (GPD) hybrid model that improves tail fit (R²_\textCDF=0.9993 vs standalone GEV’s 0.998) while maintaining architectural generalizability; (2) quantitative characterization of prompt anchoring effects across models, showing reduced dispersion but increased outliers under randomization; and (3) identification of temperature-dependent response patterns that persist across architectures, with higher temperatures amplifying length variability while preserving extreme-value mechanisms. The hybrid model’s threshold selection method enables precise verbosity control in production systems regardless of model choice. While validated on multiple architectures, generalizability to emerging model families requires further study.
%R 10.18653/v1/2025.emnlp-main.1676
%U https://aclanthology.org/2025.emnlp-main.1676/
%U https://doi.org/10.18653/v1/2025.emnlp-main.1676
%P 32992-33002

Markdown (Informal)

[Analyzing and Modeling LLM Response Lengths with Extreme Value Theory: Anchoring Effects and Hybrid Distributions](https://aclanthology.org/2025.emnlp-main.1676/) (Jiao et al., EMNLP 2025)

• Analyzing and Modeling LLM Response Lengths with Extreme Value Theory: Anchoring Effects and Hybrid Distributions (Jiao et al., EMNLP 2025)

ACL

• Liuxuan Jiao, Chen Gao, Yiqian Yang, Chenliang Zhou, YiXian Huang, Xinlei Chen, and Yong Li. 2025. Analyzing and Modeling LLM Response Lengths with Extreme Value Theory: Anchoring Effects and Hybrid Distributions. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 32992–33002, Suzhou, China. Association for Computational Linguistics.