@inproceedings{huang-etal-2026-rewardcode,
title = "{R}eward{C}ode: Training Generalist Code Reward Model via Pairwise Reinforcement Learning",
author = "Huang, Songtao and
Feng, Shiyang and
He, Xiaohan and
Yan, Xiangchao and
Zhang, Bo and
Bai, Lei",
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.448/",
pages = "9409--9431",
ISBN = "979-8-89176-462-0",
abstract = "Reward modeling aligns language models with human preferences, yet coding tasks, requiring strict logic and syntax, pose challenges due to their domain diversity. In this paper, we propose RewardCode, a generalist code reward model trained via a two-stage pipeline, enabling scalable and precise evaluation through principle-based scoring and unit test generation. Specifically, the first stage combines Structured Fine-Tuning and Rejective Fine-Tuning to distill cross-domain summarization and code evaluation capabilities from advanced LLMs. The second stage introduces Pairwise-GRPO, a reinforcement learning method that uses preference pairs to improve solution discrimination, while ensuring the generation of consistent and verifiable unit tests. Further, to train a cross-domain code reward model, we construct CodePair-18K, a dataset of verifiable code preference pairs and executable unit tests. Experiments on multiple benchmarks show that RewardCode outperforms strong baselines in evaluation accuracy, verifying its effectiveness in advancing general-purpose code LLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="huang-etal-2026-rewardcode">
<titleInfo>
<title>RewardCode: Training Generalist Code Reward Model via Pairwise Reinforcement Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Songtao</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shiyang</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaohan</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiangchao</namePart>
<namePart type="family">Yan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Bai</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>Reward modeling aligns language models with human preferences, yet coding tasks, requiring strict logic and syntax, pose challenges due to their domain diversity. In this paper, we propose RewardCode, a generalist code reward model trained via a two-stage pipeline, enabling scalable and precise evaluation through principle-based scoring and unit test generation. Specifically, the first stage combines Structured Fine-Tuning and Rejective Fine-Tuning to distill cross-domain summarization and code evaluation capabilities from advanced LLMs. The second stage introduces Pairwise-GRPO, a reinforcement learning method that uses preference pairs to improve solution discrimination, while ensuring the generation of consistent and verifiable unit tests. Further, to train a cross-domain code reward model, we construct CodePair-18K, a dataset of verifiable code preference pairs and executable unit tests. Experiments on multiple benchmarks show that RewardCode outperforms strong baselines in evaluation accuracy, verifying its effectiveness in advancing general-purpose code LLMs.</abstract>
<identifier type="citekey">huang-etal-2026-rewardcode</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.448/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>9409</start>
<end>9431</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T RewardCode: Training Generalist Code Reward Model via Pairwise Reinforcement Learning
%A Huang, Songtao
%A Feng, Shiyang
%A He, Xiaohan
%A Yan, Xiangchao
%A Zhang, Bo
%A Bai, Lei
%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 huang-etal-2026-rewardcode
%X Reward modeling aligns language models with human preferences, yet coding tasks, requiring strict logic and syntax, pose challenges due to their domain diversity. In this paper, we propose RewardCode, a generalist code reward model trained via a two-stage pipeline, enabling scalable and precise evaluation through principle-based scoring and unit test generation. Specifically, the first stage combines Structured Fine-Tuning and Rejective Fine-Tuning to distill cross-domain summarization and code evaluation capabilities from advanced LLMs. The second stage introduces Pairwise-GRPO, a reinforcement learning method that uses preference pairs to improve solution discrimination, while ensuring the generation of consistent and verifiable unit tests. Further, to train a cross-domain code reward model, we construct CodePair-18K, a dataset of verifiable code preference pairs and executable unit tests. Experiments on multiple benchmarks show that RewardCode outperforms strong baselines in evaluation accuracy, verifying its effectiveness in advancing general-purpose code LLMs.
%U https://aclanthology.org/2026.findings-emnlp.448/
%P 9409-9431
Markdown (Informal)
[RewardCode: Training Generalist Code Reward Model via Pairwise Reinforcement Learning](https://aclanthology.org/2026.findings-emnlp.448/) (Huang et al., Findings 2026)
ACL