<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2026-aster">
    <titleInfo>
        <title>ASTER: Agentic Scaling with Tool-integrated Extended Reasoning</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Xuqin</namePart>
        <namePart type="family">Zhang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Quan</namePart>
        <namePart type="family">He</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Zhenrui</namePart>
        <namePart type="family">Zheng</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Zongzhang</namePart>
        <namePart type="family">Zhang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Xu</namePart>
        <namePart type="family">He</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Dong</namePart>
        <namePart type="family">Li</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>Reinforcement Learning (RL) has emerged as a dominant paradigm for eliciting long-horizon reasoning in Large Language Models (LLMs). However, scaling Tool-Integrated Reasoning (TIR) via RL remains challenging due to interaction collapse: a pathological state where models fail to sustain multi-turn tool usage, instead degenerating into heavy internal reasoning with only trivial, post-hoc code verification. We systematically study three questions: (i) how cold-start Supervised Fine-Tuning (SFT) induces an agentic, tool-using behavioral prior, (ii) how teacher-trajectory structure and interaction density shape exploration and downstream RL outcomes, and (iii) how the RL interaction budget affects learning dynamics and generalization under varying inference-time budgets. We then introduce ASTER (Agentic Scaling with Tool-integrated Extended Reasoning), a framework that circumvents this collapse through a targeted cold-start strategy prioritizing structured, interaction-dense trajectories. We find that a small expert cold-start set of just 4K structurally rich, interaction-dense trajectories yields the strongest downstream performance, establishing a robust prior that enables superior exploration during extended RL training. Extensive evaluations demonstrate that ASTER-4B achieves strong results on competitive mathematical benchmarks, reaching 89.6% mean accuracy on AIME2025 and remaining competitive with leading frontier open-source models. See more details at https://github.com/CodeByLeo-bit/ASTER.</abstract>
    <identifier type="citekey">zhang-etal-2026-aster</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-emnlp.1704/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>36427</start>
            <end>36451</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T ASTER: Agentic Scaling with Tool-integrated Extended Reasoning
%A Zhang, Xuqin
%A He, Quan
%A Zheng, Zhenrui
%A Zhang, Zongzhang
%A He, Xu
%A Li, Dong
%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 zhang-etal-2026-aster
%X Reinforcement Learning (RL) has emerged as a dominant paradigm for eliciting long-horizon reasoning in Large Language Models (LLMs). However, scaling Tool-Integrated Reasoning (TIR) via RL remains challenging due to interaction collapse: a pathological state where models fail to sustain multi-turn tool usage, instead degenerating into heavy internal reasoning with only trivial, post-hoc code verification. We systematically study three questions: (i) how cold-start Supervised Fine-Tuning (SFT) induces an agentic, tool-using behavioral prior, (ii) how teacher-trajectory structure and interaction density shape exploration and downstream RL outcomes, and (iii) how the RL interaction budget affects learning dynamics and generalization under varying inference-time budgets. We then introduce ASTER (Agentic Scaling with Tool-integrated Extended Reasoning), a framework that circumvents this collapse through a targeted cold-start strategy prioritizing structured, interaction-dense trajectories. We find that a small expert cold-start set of just 4K structurally rich, interaction-dense trajectories yields the strongest downstream performance, establishing a robust prior that enables superior exploration during extended RL training. Extensive evaluations demonstrate that ASTER-4B achieves strong results on competitive mathematical benchmarks, reaching 89.6% mean accuracy on AIME2025 and remaining competitive with leading frontier open-source models. See more details at https://github.com/CodeByLeo-bit/ASTER.
%U https://aclanthology.org/2026.findings-emnlp.1704/
%P 36427-36451

Markdown (Informal)

[ASTER: Agentic Scaling with Tool-integrated Extended Reasoning](https://aclanthology.org/2026.findings-emnlp.1704/) (Zhang et al., Findings 2026)

• ASTER: Agentic Scaling with Tool-integrated Extended Reasoning (Zhang et al., Findings 2026)

ACL

• Xuqin Zhang, Quan He, Zhenrui Zheng, Zongzhang Zhang, Xu He, and Dong Li. 2026. ASTER: Agentic Scaling with Tool-integrated Extended Reasoning. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 36427–36451, Budapest, Hungary. Association for Computational Linguistics.