<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="li-etal-2026-target">
    <titleInfo>
        <title>Target-Local Verifier Choice in Best-of-K Reasoning Selection</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Xin</namePart>
        <namePart type="family">Li</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Hao</namePart>
        <namePart type="family">Jiang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Weisi</namePart>
        <namePart type="family">Lin</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-10</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Hung-Yi</namePart>
            <namePart type="family">Lee</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Sunipa</namePart>
            <namePart type="family">Dev</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jessy</namePart>
            <namePart type="family">Li</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Desmond</namePart>
            <namePart type="family">Elliott</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Budapest, Hungary</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-462-0</identifier>
    </relatedItem>
    <abstract>Modern reasoning systems increasingly rely on verifier-guided selection: sample several solution traces, score them with a reward model, and return the top-ranked answer. In multi-generator deployments, this verifier is often reused across generators. We ask whether this reuse assumption is reliable. On a 34-generator math panel with K=64 candidates per problem and 7 verifiers, the 7B process reward model (PRM-7B) is the strongest fixed verifier, but a per-target single-verifier oracle gains +1.6 percentage points (pp). We then ask whether this target-local mismatch is actionable. Four label-free candidate-text statistics predict the coarse PRM outcome-verifier regime, and three budgeted responses improve over fixed PRM on the 22-target first-pass panel: a no-label gated router (+1.41 pp), PRM+V4 fusion (+2.35 pp), and a labelled PRM-only verifier-allocation arbiter (PVA-Adapt; +2.26 pp). A same-panel American Invitational Mathematics Examination (AIME) 2026 held-out slice gives +3.88 pp, and a second candidate-bank check keeps the scored and labelled gains positive while weakening the no-label router. We frame the result as a seed- and verifier-family-conditioned deployment diagnostic rather than a universal routing rule; a supervised fine-tuning (SFT) replication also finds no positive transfer from better selected traces.</abstract>
    <identifier type="citekey">li-etal-2026-target</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-emnlp.374/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>7892</start>
            <end>7915</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Target-Local Verifier Choice in Best-of-K Reasoning Selection
%A Li, Xin
%A Jiang, Hao
%A Lin, Weisi
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F li-etal-2026-target
%X Modern reasoning systems increasingly rely on verifier-guided selection: sample several solution traces, score them with a reward model, and return the top-ranked answer. In multi-generator deployments, this verifier is often reused across generators. We ask whether this reuse assumption is reliable. On a 34-generator math panel with K=64 candidates per problem and 7 verifiers, the 7B process reward model (PRM-7B) is the strongest fixed verifier, but a per-target single-verifier oracle gains +1.6 percentage points (pp). We then ask whether this target-local mismatch is actionable. Four label-free candidate-text statistics predict the coarse PRM outcome-verifier regime, and three budgeted responses improve over fixed PRM on the 22-target first-pass panel: a no-label gated router (+1.41 pp), PRM+V4 fusion (+2.35 pp), and a labelled PRM-only verifier-allocation arbiter (PVA-Adapt; +2.26 pp). A same-panel American Invitational Mathematics Examination (AIME) 2026 held-out slice gives +3.88 pp, and a second candidate-bank check keeps the scored and labelled gains positive while weakening the no-label router. We frame the result as a seed- and verifier-family-conditioned deployment diagnostic rather than a universal routing rule; a supervised fine-tuning (SFT) replication also finds no positive transfer from better selected traces.
%U https://aclanthology.org/2026.findings-emnlp.374/
%P 7892-7915

Markdown (Informal)

[Target-Local Verifier Choice in Best-of-K Reasoning Selection](https://aclanthology.org/2026.findings-emnlp.374/) (Li et al., Findings 2026)

• Target-Local Verifier Choice in Best-of-K Reasoning Selection (Li et al., Findings 2026)

ACL

• Xin Li, Hao Jiang, and Weisi Lin. 2026. Target-Local Verifier Choice in Best-of-K Reasoning Selection. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 7892–7915, Budapest, Hungary. Association for Computational Linguistics.