<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chen-etal-2025-magicore">
    <titleInfo>
        <title>MAgICoRe: Multi-Agent, Iterative, Coarse-to-Fine Refinement for Reasoning</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Justin</namePart>
        <namePart type="family">Chen</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Archiki</namePart>
        <namePart type="family">Prasad</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Swarnadeep</namePart>
        <namePart type="family">Saha</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Elias</namePart>
        <namePart type="family">Stengel-Eskin</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Mohit</namePart>
        <namePart type="family">Bansal</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>Large language model (LLM) reasoning can be improved by scaling test-time compute with aggregation, i.e., generating multiple samples and aggregating over them. While improving performance, this strategy often reaches a saturation point beyond which additional compute provides no return. Refinement offers an alternative by using model-generated feedback to improve answer quality. However, refinement faces three key challenges: (1) Excessive refinement: Uniformly refining all instances can cause over-correction and reduce overall performance. (2) Inability to localize and address errors: LLMs struggle to identify and correct their own mistakes. (3) Insufficient refinement: Stopping refinement too soon could leave errors unaddressed. To tackle these issues, we propose MAgICoRe, a framework for Multi-Agent Iteration for Coarse-to-fine Refinement. MAgICoRe mitigates excessive refinement by categorizing problems as easy or hard, solving easy problems with coarse-grained aggregation, and solving the hard ones with fine-grained multi-agent refinement. To better localize errors, we incorporate external step-wise reward model scores, and to ensure sufficient refinement, we iteratively refine the solutions using a multi-agent setup. We evaluate MAgICoRe on Llama-3-8B and GPT- 3.5 and show its effectiveness across seven reasoning datasets. One iteration of MAgICoRe beats Self-Consistency by 3.4%, Best-of-k by 3.2%, and Self-Refine by 4.0% even when these baselines use k = 120, and MAgICoRe uses less than 50% of the compute.</abstract>
    <identifier type="citekey">chen-etal-2025-magicore</identifier>
    <location>
        <url>https://aclanthology.org/2025.emnlp-main.1660/</url>
    </location>
    <part>
        <date>2025-11</date>
        <extent unit="page">
            <start>32651</start>
            <end>32674</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T MAgICoRe: Multi-Agent, Iterative, Coarse-to-Fine Refinement for Reasoning
%A Chen, Justin
%A Prasad, Archiki
%A Saha, Swarnadeep
%A Stengel-Eskin, Elias
%A Bansal, Mohit
%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 chen-etal-2025-magicore
%X Large language model (LLM) reasoning can be improved by scaling test-time compute with aggregation, i.e., generating multiple samples and aggregating over them. While improving performance, this strategy often reaches a saturation point beyond which additional compute provides no return. Refinement offers an alternative by using model-generated feedback to improve answer quality. However, refinement faces three key challenges: (1) Excessive refinement: Uniformly refining all instances can cause over-correction and reduce overall performance. (2) Inability to localize and address errors: LLMs struggle to identify and correct their own mistakes. (3) Insufficient refinement: Stopping refinement too soon could leave errors unaddressed. To tackle these issues, we propose MAgICoRe, a framework for Multi-Agent Iteration for Coarse-to-fine Refinement. MAgICoRe mitigates excessive refinement by categorizing problems as easy or hard, solving easy problems with coarse-grained aggregation, and solving the hard ones with fine-grained multi-agent refinement. To better localize errors, we incorporate external step-wise reward model scores, and to ensure sufficient refinement, we iteratively refine the solutions using a multi-agent setup. We evaluate MAgICoRe on Llama-3-8B and GPT- 3.5 and show its effectiveness across seven reasoning datasets. One iteration of MAgICoRe beats Self-Consistency by 3.4%, Best-of-k by 3.2%, and Self-Refine by 4.0% even when these baselines use k = 120, and MAgICoRe uses less than 50% of the compute.
%U https://aclanthology.org/2025.emnlp-main.1660/
%P 32651-32674

Markdown (Informal)

[MAgICoRe: Multi-Agent, Iterative, Coarse-to-Fine Refinement for Reasoning](https://aclanthology.org/2025.emnlp-main.1660/) (Chen et al., EMNLP 2025)

• MAgICoRe: Multi-Agent, Iterative, Coarse-to-Fine Refinement for Reasoning (Chen et al., EMNLP 2025)

ACL

• Justin Chen, Archiki Prasad, Swarnadeep Saha, Elias Stengel-Eskin, and Mohit Bansal. 2025. MAgICoRe: Multi-Agent, Iterative, Coarse-to-Fine Refinement for Reasoning. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 32651–32674, Suzhou, China. Association for Computational Linguistics.