<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lee-etal-2026-efficient">
    <titleInfo>
        <title>Efficient Process Reward Modeling via Contrastive Mutual Information</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Nakyung</namePart>
        <namePart type="family">Lee</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Sangwoo</namePart>
        <namePart type="family">Hong</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jungwoo</namePart>
        <namePart type="family">Lee</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>Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)</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">979-8-89176-390-6</identifier>
    </relatedItem>
    <abstract>Recent research has devoted considerable effort to verifying the intermediate reasoning steps of chain-of-thought (CoT) trajectories using process reward models (PRMs) and other verifier models. However, training a PRM typically requires human annotators to assign reward scores to each reasoning step, which is both costly and time-consuming. Existing automated approaches, such as Monte Carlo (MC) estimation, also demand substantial computational resources due to repeated LLM rollouts. To overcome these limitations, we propose **contrastive pointwise mutual information (CPMI)**, a novel automatic reward labeling method that leverages the model’s internal probability to infer step-level supervision while significantly reducing the computational burden of annotating dataset. CPMI quantifies how much a reasoning step increases the **mutual information** between the step and the correct target answer relative to hard-negative alternatives. This **contrastive** signal serves as a proxy for the step’s contribution to the final solution and yields a reliable reward. The experimental results show that CPMI-based labeling reduces dataset construction time by **84%** and token generation by **98%** compared to MC estimation, while achieving higher accuracy on process-level evaluations and mathematical reasoning benchmarks.</abstract>
    <identifier type="citekey">lee-etal-2026-efficient</identifier>
    <identifier type="doi">10.18653/v1/2026.acl-long.1744</identifier>
    <location>
        <url>https://aclanthology.org/2026.acl-long.1744/</url>
    </location>
    <part>
        <date>2026-07</date>
        <extent unit="page">
            <start>37603</start>
            <end>37617</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Efficient Process Reward Modeling via Contrastive Mutual Information
%A Lee, Nakyung
%A Hong, Sangwoo
%A Lee, Jungwoo
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-390-6
%F lee-etal-2026-efficient
%X Recent research has devoted considerable effort to verifying the intermediate reasoning steps of chain-of-thought (CoT) trajectories using process reward models (PRMs) and other verifier models. However, training a PRM typically requires human annotators to assign reward scores to each reasoning step, which is both costly and time-consuming. Existing automated approaches, such as Monte Carlo (MC) estimation, also demand substantial computational resources due to repeated LLM rollouts. To overcome these limitations, we propose **contrastive pointwise mutual information (CPMI)**, a novel automatic reward labeling method that leverages the model’s internal probability to infer step-level supervision while significantly reducing the computational burden of annotating dataset. CPMI quantifies how much a reasoning step increases the **mutual information** between the step and the correct target answer relative to hard-negative alternatives. This **contrastive** signal serves as a proxy for the step’s contribution to the final solution and yields a reliable reward. The experimental results show that CPMI-based labeling reduces dataset construction time by **84%** and token generation by **98%** compared to MC estimation, while achieving higher accuracy on process-level evaluations and mathematical reasoning benchmarks.
%R 10.18653/v1/2026.acl-long.1744
%U https://aclanthology.org/2026.acl-long.1744/
%U https://doi.org/10.18653/v1/2026.acl-long.1744
%P 37603-37617

Markdown (Informal)

[Efficient Process Reward Modeling via Contrastive Mutual Information](https://aclanthology.org/2026.acl-long.1744/) (Lee et al., ACL 2026)

• Efficient Process Reward Modeling via Contrastive Mutual Information (Lee et al., ACL 2026)

ACL

• Nakyung Lee, Sangwoo Hong, and Jungwoo Lee. 2026. Efficient Process Reward Modeling via Contrastive Mutual Information. In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 37603–37617, San Diego, California, United States. Association for Computational Linguistics.