@inproceedings{li-etal-2026-ceprm,
title = "{C}e{PRM}: Generative Process Reward Modeling with Calibrated Values and Repair-Aware Supervision",
author = "Li, Zongxuan and
Dong, Yixiang and
Hu, Bo and
Cao, Gonghao and
Chen, Yanzhen and
Liu, Huan",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1923/",
pages = "40991--41007",
ISBN = "979-8-89176-462-0",
abstract = "Generative process reward models (PRMs) improve step-level verification of mathematical reasoning, yet their effectiveness is limited by the quality of process supervision. The dominant Monte Carlo (MC) estimation derives step labels from final-answer accuracy rather than semantic correctness, producing systematic label noise, while standard pipelines discard the implicit tree structure of rollout data that encodes informative failure patterns. We propose CePRM, which addresses both limitations through Step-Cognitive MC Value Scaling, calibrating raw MC scores with step-level semantic signals via multiplicative adjustment, and pivot-step-directed repair mining, which exploits value cliffs in failed trajectories to construct same-prefix contrastive correction pairs via Sibling Advantage filtering. Experiments on ProcessBench and PRMBench show that CePRM, with a 4B backbone, achieves competitive or superior error localization and process verification against 7B-scale baselines, with notably strong error-side recognition."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="li-etal-2026-ceprm">
<titleInfo>
<title>CePRM: Generative Process Reward Modeling with Calibrated Values and Repair-Aware Supervision</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zongxuan</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yixiang</namePart>
<namePart type="family">Dong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gonghao</namePart>
<namePart type="family">Cao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanzhen</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Huan</namePart>
<namePart type="family">Liu</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>Generative process reward models (PRMs) improve step-level verification of mathematical reasoning, yet their effectiveness is limited by the quality of process supervision. The dominant Monte Carlo (MC) estimation derives step labels from final-answer accuracy rather than semantic correctness, producing systematic label noise, while standard pipelines discard the implicit tree structure of rollout data that encodes informative failure patterns. We propose CePRM, which addresses both limitations through Step-Cognitive MC Value Scaling, calibrating raw MC scores with step-level semantic signals via multiplicative adjustment, and pivot-step-directed repair mining, which exploits value cliffs in failed trajectories to construct same-prefix contrastive correction pairs via Sibling Advantage filtering. Experiments on ProcessBench and PRMBench show that CePRM, with a 4B backbone, achieves competitive or superior error localization and process verification against 7B-scale baselines, with notably strong error-side recognition.</abstract>
<identifier type="citekey">li-etal-2026-ceprm</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1923/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>40991</start>
<end>41007</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T CePRM: Generative Process Reward Modeling with Calibrated Values and Repair-Aware Supervision
%A Li, Zongxuan
%A Dong, Yixiang
%A Hu, Bo
%A Cao, Gonghao
%A Chen, Yanzhen
%A Liu, Huan
%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-ceprm
%X Generative process reward models (PRMs) improve step-level verification of mathematical reasoning, yet their effectiveness is limited by the quality of process supervision. The dominant Monte Carlo (MC) estimation derives step labels from final-answer accuracy rather than semantic correctness, producing systematic label noise, while standard pipelines discard the implicit tree structure of rollout data that encodes informative failure patterns. We propose CePRM, which addresses both limitations through Step-Cognitive MC Value Scaling, calibrating raw MC scores with step-level semantic signals via multiplicative adjustment, and pivot-step-directed repair mining, which exploits value cliffs in failed trajectories to construct same-prefix contrastive correction pairs via Sibling Advantage filtering. Experiments on ProcessBench and PRMBench show that CePRM, with a 4B backbone, achieves competitive or superior error localization and process verification against 7B-scale baselines, with notably strong error-side recognition.
%U https://aclanthology.org/2026.findings-emnlp.1923/
%P 40991-41007
Markdown (Informal)
[CePRM: Generative Process Reward Modeling with Calibrated Values and Repair-Aware Supervision](https://aclanthology.org/2026.findings-emnlp.1923/) (Li et al., Findings 2026)
ACL