@inproceedings{zhang-etal-2025-dorm,
title = "{DORM}: Preference Data Weights Optimization for Reward Modeling in {LLM} Alignment",
author = "Zhang, Rongzhi and
Zhang, Chenwei and
Zhang, Xinyang and
Qiu, Liang and
Jiang, Haoming and
Zhuang, Yuchen and
Zhang, Qingru and
Yun, Hyokun and
Li, Xian and
Yin, Bing and
Zhao, Tuo and
Zhang, Chao",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-emnlp.1237/",
pages = "22721--22739",
ISBN = "979-8-89176-335-7",
abstract = "Aligning large language models (LLMs) with human preferences relies heavily on high-quality reward models. However, existing approaches struggle with two critical challenges: noisy preference labels and the varying importance of preference samples. We introduce DORM, a method that enhances reward modeling by learning to dynamically weigh preference data.DORM initializes data importance using a combination of model uncertainty and prediction disagreement, then iteratively refines them via bilevel optimization to maximize validation performance. Using only 50k samples, DORM trains a 12B reward model that achieves 90.5{\%} accuracy on RewardBench, matching the performance of models trained on significantly larger datasets. Furthermore, downstream alignment tasks show that fine-tuned LLMs with DORM achieve a 61.2{\%} win rate against baseline methods, highlighting its data efficiency and generalizability."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2025-dorm">
<titleInfo>
<title>DORM: Preference Data Weights Optimization for Reward Modeling in LLM Alignment</title>
</titleInfo>
<name type="personal">
<namePart type="given">Rongzhi</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chenwei</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinyang</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Liang</namePart>
<namePart type="family">Qiu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haoming</namePart>
<namePart type="family">Jiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuchen</namePart>
<namePart type="family">Zhuang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qingru</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hyokun</namePart>
<namePart type="family">Yun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xian</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bing</namePart>
<namePart type="family">Yin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tuo</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chao</namePart>
<namePart type="family">Zhang</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>Findings of the Association for Computational Linguistics: EMNLP 2025</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">979-8-89176-335-7</identifier>
</relatedItem>
<abstract>Aligning large language models (LLMs) with human preferences relies heavily on high-quality reward models. However, existing approaches struggle with two critical challenges: noisy preference labels and the varying importance of preference samples. We introduce DORM, a method that enhances reward modeling by learning to dynamically weigh preference data.DORM initializes data importance using a combination of model uncertainty and prediction disagreement, then iteratively refines them via bilevel optimization to maximize validation performance. Using only 50k samples, DORM trains a 12B reward model that achieves 90.5% accuracy on RewardBench, matching the performance of models trained on significantly larger datasets. Furthermore, downstream alignment tasks show that fine-tuned LLMs with DORM achieve a 61.2% win rate against baseline methods, highlighting its data efficiency and generalizability.</abstract>
<identifier type="citekey">zhang-etal-2025-dorm</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.1237/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>22721</start>
<end>22739</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T DORM: Preference Data Weights Optimization for Reward Modeling in LLM Alignment
%A Zhang, Rongzhi
%A Zhang, Chenwei
%A Zhang, Xinyang
%A Qiu, Liang
%A Jiang, Haoming
%A Zhuang, Yuchen
%A Zhang, Qingru
%A Yun, Hyokun
%A Li, Xian
%A Yin, Bing
%A Zhao, Tuo
%A Zhang, Chao
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Findings of the Association for Computational Linguistics: EMNLP 2025
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-335-7
%F zhang-etal-2025-dorm
%X Aligning large language models (LLMs) with human preferences relies heavily on high-quality reward models. However, existing approaches struggle with two critical challenges: noisy preference labels and the varying importance of preference samples. We introduce DORM, a method that enhances reward modeling by learning to dynamically weigh preference data.DORM initializes data importance using a combination of model uncertainty and prediction disagreement, then iteratively refines them via bilevel optimization to maximize validation performance. Using only 50k samples, DORM trains a 12B reward model that achieves 90.5% accuracy on RewardBench, matching the performance of models trained on significantly larger datasets. Furthermore, downstream alignment tasks show that fine-tuned LLMs with DORM achieve a 61.2% win rate against baseline methods, highlighting its data efficiency and generalizability.
%U https://aclanthology.org/2025.findings-emnlp.1237/
%P 22721-22739
Markdown (Informal)
[DORM: Preference Data Weights Optimization for Reward Modeling in LLM Alignment](https://aclanthology.org/2025.findings-emnlp.1237/) (Zhang et al., Findings 2025)
ACL
- Rongzhi Zhang, Chenwei Zhang, Xinyang Zhang, Liang Qiu, Haoming Jiang, Yuchen Zhuang, Qingru Zhang, Hyokun Yun, Xian Li, Bing Yin, Tuo Zhao, and Chao Zhang. 2025. DORM: Preference Data Weights Optimization for Reward Modeling in LLM Alignment. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 22721–22739, Suzhou, China. Association for Computational Linguistics.