AI/ML News & Innovations Hub

@inproceedings{jang-etal-2026-bad,
    title = "A Few Bad Apples Spoil the Bunch: Preventing Global Entropy Collapse Driven by a Small Set of Tokens in {LLM} Reasoning",
    author = "Jang, Jaeeun  and
      Lee, Hansle  and
      Kim, Sangmin",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-acl.641/",
    pages = "13134--13154",
    ISBN = "979-8-89176-395-1",
    abstract = "Reinforcement Learning with Verifiable Rewards (RLVR) and Reinforcement Learning from Internal Feedback (RLIF) often fail to benefit from test-time compute due to entropy collapse and the resulting loss of reasoning diversity. We show that this collapse is driven not by uniform entropy decay, but by premature overconfidence at a small number of structurally critical decision points. Based on a token-level analysis of GRPO-style policy optimization, we propose SCOPE (Structural Collapse-aware Optimization via Partial Entropy control), which assigns each generated token a redistribution score and applies selective KL regularization to only the top $\sim 5${\%} of tokens under this score. Across model scales and architectures on math reasoning benchmarks, SCOPE consistently improves performance under both RLVR and RLIF settings, demonstrating that targeted entropy control at a vanishingly small subset of tokens is sufficient to sustain reasoning diversity and effective test-time scaling."
}

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jang-etal-2026-bad">
    <titleInfo>
        <title>A Few Bad Apples Spoil the Bunch: Preventing Global Entropy Collapse Driven by a Small Set of Tokens in LLM Reasoning</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Jaeeun</namePart>
        <namePart type="family">Jang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Hansle</namePart>
        <namePart type="family">Lee</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Sangmin</namePart>
        <namePart type="family">Kim</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>Findings of the Association for Computational Linguistics: ACL 2026</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Maria</namePart>
            <namePart type="family">Liakata</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Viviane</namePart>
            <namePart type="given">P</namePart>
            <namePart type="family">Moreira</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jiajun</namePart>
            <namePart type="family">Zhang</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">David</namePart>
            <namePart type="family">Jurgens</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, United States</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-395-1</identifier>
    </relatedItem>
    <abstract>Reinforcement Learning with Verifiable Rewards (RLVR) and Reinforcement Learning from Internal Feedback (RLIF) often fail to benefit from test-time compute due to entropy collapse and the resulting loss of reasoning diversity. We show that this collapse is driven not by uniform entropy decay, but by premature overconfidence at a small number of structurally critical decision points. Based on a token-level analysis of GRPO-style policy optimization, we propose SCOPE (Structural Collapse-aware Optimization via Partial Entropy control), which assigns each generated token a redistribution score and applies selective KL regularization to only the top \sim 5% of tokens under this score. Across model scales and architectures on math reasoning benchmarks, SCOPE consistently improves performance under both RLVR and RLIF settings, demonstrating that targeted entropy control at a vanishingly small subset of tokens is sufficient to sustain reasoning diversity and effective test-time scaling.</abstract>
    <identifier type="citekey">jang-etal-2026-bad</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-acl.641/</url>
    </location>
    <part>
        <date>2026-07</date>
        <extent unit="page">
            <start>13134</start>
            <end>13154</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T A Few Bad Apples Spoil the Bunch: Preventing Global Entropy Collapse Driven by a Small Set of Tokens in LLM Reasoning
%A Jang, Jaeeun
%A Lee, Hansle
%A Kim, Sangmin
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F jang-etal-2026-bad
%X Reinforcement Learning with Verifiable Rewards (RLVR) and Reinforcement Learning from Internal Feedback (RLIF) often fail to benefit from test-time compute due to entropy collapse and the resulting loss of reasoning diversity. We show that this collapse is driven not by uniform entropy decay, but by premature overconfidence at a small number of structurally critical decision points. Based on a token-level analysis of GRPO-style policy optimization, we propose SCOPE (Structural Collapse-aware Optimization via Partial Entropy control), which assigns each generated token a redistribution score and applies selective KL regularization to only the top \sim 5% of tokens under this score. Across model scales and architectures on math reasoning benchmarks, SCOPE consistently improves performance under both RLVR and RLIF settings, demonstrating that targeted entropy control at a vanishingly small subset of tokens is sufficient to sustain reasoning diversity and effective test-time scaling.
%U https://aclanthology.org/2026.findings-acl.641/
%P 13134-13154

Markdown (Informal)

[A Few Bad Apples Spoil the Bunch: Preventing Global Entropy Collapse Driven by a Small Set of Tokens in LLM Reasoning](https://aclanthology.org/2026.findings-acl.641/) (Jang et al., Findings 2026)