@inproceedings{yi-etal-2025-sppd,
title = "{SPPD}: Self-training with Process Preference Learning Using Dynamic Value Margin",
author = "Yi, Hao and
Li, Qingyang and
Hu, Yulan and
Zhang, Fuzheng and
Zhang, Di and
Liu, Yong",
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.19/",
pages = "324--337",
ISBN = "979-8-89176-335-7",
abstract = "Enhancing the numerical and logical reasoning capabilities of Large Language Models (LLMs) has become a prominent research focus. Existing approaches exhibit notable limitations: inference-phase techniques, such as Chain of Thought, depend on prompt engineering and pretrained knowledge; sentence-level Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) struggle to ensure step-wise mathematical correctness and often rely on model distillation or human annotations; Reinforcement Learning (RL) methods entail high GPU memory consumption and training instability. To overcome these challenges, we propose \textbf{S}elf-training with \textbf{P}rocess \textbf{P}reference learning using \textbf{D}ynamic value margin (\textbf{SPPD}). SPPD formulates reasoning as a process-based Markov Decision Process (MDP), leveraging the Bellman optimality equation to derive a \textbf{dynamic value margin} for step-level preference optimization. It further incorporates tree-based self-sampling of model responses, \textbf{eliminating the need for distillation.} We theoretically establish that SPPD is \textbf{equivalent to on-policy policy gradient methods} under constrained reward functions. Experimental results on 7B-scale models show consistent superiority across both in-domain and out-of-domain mathematical benchmarks."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yi-etal-2025-sppd">
<titleInfo>
<title>SPPD: Self-training with Process Preference Learning Using Dynamic Value Margin</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hao</namePart>
<namePart type="family">Yi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qingyang</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yulan</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fuzheng</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Di</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yong</namePart>
<namePart type="family">Liu</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>Enhancing the numerical and logical reasoning capabilities of Large Language Models (LLMs) has become a prominent research focus. Existing approaches exhibit notable limitations: inference-phase techniques, such as Chain of Thought, depend on prompt engineering and pretrained knowledge; sentence-level Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) struggle to ensure step-wise mathematical correctness and often rely on model distillation or human annotations; Reinforcement Learning (RL) methods entail high GPU memory consumption and training instability. To overcome these challenges, we propose Self-training with Process Preference learning using Dynamic value margin (SPPD). SPPD formulates reasoning as a process-based Markov Decision Process (MDP), leveraging the Bellman optimality equation to derive a dynamic value margin for step-level preference optimization. It further incorporates tree-based self-sampling of model responses, eliminating the need for distillation. We theoretically establish that SPPD is equivalent to on-policy policy gradient methods under constrained reward functions. Experimental results on 7B-scale models show consistent superiority across both in-domain and out-of-domain mathematical benchmarks.</abstract>
<identifier type="citekey">yi-etal-2025-sppd</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.19/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>324</start>
<end>337</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SPPD: Self-training with Process Preference Learning Using Dynamic Value Margin
%A Yi, Hao
%A Li, Qingyang
%A Hu, Yulan
%A Zhang, Fuzheng
%A Zhang, Di
%A Liu, Yong
%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 yi-etal-2025-sppd
%X Enhancing the numerical and logical reasoning capabilities of Large Language Models (LLMs) has become a prominent research focus. Existing approaches exhibit notable limitations: inference-phase techniques, such as Chain of Thought, depend on prompt engineering and pretrained knowledge; sentence-level Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) struggle to ensure step-wise mathematical correctness and often rely on model distillation or human annotations; Reinforcement Learning (RL) methods entail high GPU memory consumption and training instability. To overcome these challenges, we propose Self-training with Process Preference learning using Dynamic value margin (SPPD). SPPD formulates reasoning as a process-based Markov Decision Process (MDP), leveraging the Bellman optimality equation to derive a dynamic value margin for step-level preference optimization. It further incorporates tree-based self-sampling of model responses, eliminating the need for distillation. We theoretically establish that SPPD is equivalent to on-policy policy gradient methods under constrained reward functions. Experimental results on 7B-scale models show consistent superiority across both in-domain and out-of-domain mathematical benchmarks.
%U https://aclanthology.org/2025.findings-emnlp.19/
%P 324-337
Markdown (Informal)
[SPPD: Self-training with Process Preference Learning Using Dynamic Value Margin](https://aclanthology.org/2025.findings-emnlp.19/) (Yi et al., Findings 2025)
ACL