@inproceedings{li-etal-2025-token,
title = "Token-level Preference Self-Alignment Optimization for Multi-style Outline Controllable Generation",
author = "Li, Zihao and
Xu, Xuekong and
Chen, Ziyao and
Zou, Lixin and
Wu, Haijun and
Chen, Qiang and
Li, Chenliang",
editor = "Che, Wanxiang and
Nabende, Joyce and
Shutova, Ekaterina and
Pilehvar, Mohammad Taher",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2025",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-acl.823/",
doi = "10.18653/v1/2025.findings-acl.823",
pages = "15974--16007",
ISBN = "979-8-89176-256-5",
abstract = "Multi-style outline controllable generation is crucial for multiple applications, including document semantic structuring and retrieval-augmented generation.The great success of preference alignment approaches encourages their application in controllable generation tasks.However, these attempts encounter several limitations: (1) response pair requirements, (2) substantial computation costs, and (3) insufficient exploitation of fine-grained preference signals.To address these problems, we propose a token-level preference self-alignment optimization, named TKPO, for outline controllable generation. TKPO extends the Bradley-Terry model from pair-wise to list-wise comparison, which is further applied at the token level for fine-grained preference signal utilization. In comparison to the representative methods, e.g., DPO, TKPO does not require response pairs; instead, we propose a controllable attributes-driven method to construct reject samples for self-alignment. Additionally, TKPO optimizes only the base model, thereby avoiding additional memory usage and substantial computational costs.We curate two outline controllable generation datasets with regard to language style and level-of-detail.Extensive experiments demonstrate that TKPO outperforms DPO by up to 19.28{\%} in performance while requiring only 56.25{\%} in training time.We release the code and datasets resources at https://github.com/WHUIR/TKPO."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="li-etal-2025-token">
<titleInfo>
<title>Token-level Preference Self-Alignment Optimization for Multi-style Outline Controllable Generation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zihao</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xuekong</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ziyao</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lixin</namePart>
<namePart type="family">Zou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haijun</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qiang</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chenliang</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wanxiang</namePart>
<namePart type="family">Che</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joyce</namePart>
<namePart type="family">Nabende</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ekaterina</namePart>
<namePart type="family">Shutova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohammad</namePart>
<namePart type="given">Taher</namePart>
<namePart type="family">Pilehvar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Vienna, Austria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-256-5</identifier>
</relatedItem>
<abstract>Multi-style outline controllable generation is crucial for multiple applications, including document semantic structuring and retrieval-augmented generation.The great success of preference alignment approaches encourages their application in controllable generation tasks.However, these attempts encounter several limitations: (1) response pair requirements, (2) substantial computation costs, and (3) insufficient exploitation of fine-grained preference signals.To address these problems, we propose a token-level preference self-alignment optimization, named TKPO, for outline controllable generation. TKPO extends the Bradley-Terry model from pair-wise to list-wise comparison, which is further applied at the token level for fine-grained preference signal utilization. In comparison to the representative methods, e.g., DPO, TKPO does not require response pairs; instead, we propose a controllable attributes-driven method to construct reject samples for self-alignment. Additionally, TKPO optimizes only the base model, thereby avoiding additional memory usage and substantial computational costs.We curate two outline controllable generation datasets with regard to language style and level-of-detail.Extensive experiments demonstrate that TKPO outperforms DPO by up to 19.28% in performance while requiring only 56.25% in training time.We release the code and datasets resources at https://github.com/WHUIR/TKPO.</abstract>
<identifier type="citekey">li-etal-2025-token</identifier>
<identifier type="doi">10.18653/v1/2025.findings-acl.823</identifier>
<location>
<url>https://aclanthology.org/2025.findings-acl.823/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>15974</start>
<end>16007</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Token-level Preference Self-Alignment Optimization for Multi-style Outline Controllable Generation
%A Li, Zihao
%A Xu, Xuekong
%A Chen, Ziyao
%A Zou, Lixin
%A Wu, Haijun
%A Chen, Qiang
%A Li, Chenliang
%Y Che, Wanxiang
%Y Nabende, Joyce
%Y Shutova, Ekaterina
%Y Pilehvar, Mohammad Taher
%S Findings of the Association for Computational Linguistics: ACL 2025
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-256-5
%F li-etal-2025-token
%X Multi-style outline controllable generation is crucial for multiple applications, including document semantic structuring and retrieval-augmented generation.The great success of preference alignment approaches encourages their application in controllable generation tasks.However, these attempts encounter several limitations: (1) response pair requirements, (2) substantial computation costs, and (3) insufficient exploitation of fine-grained preference signals.To address these problems, we propose a token-level preference self-alignment optimization, named TKPO, for outline controllable generation. TKPO extends the Bradley-Terry model from pair-wise to list-wise comparison, which is further applied at the token level for fine-grained preference signal utilization. In comparison to the representative methods, e.g., DPO, TKPO does not require response pairs; instead, we propose a controllable attributes-driven method to construct reject samples for self-alignment. Additionally, TKPO optimizes only the base model, thereby avoiding additional memory usage and substantial computational costs.We curate two outline controllable generation datasets with regard to language style and level-of-detail.Extensive experiments demonstrate that TKPO outperforms DPO by up to 19.28% in performance while requiring only 56.25% in training time.We release the code and datasets resources at https://github.com/WHUIR/TKPO.
%R 10.18653/v1/2025.findings-acl.823
%U https://aclanthology.org/2025.findings-acl.823/
%U https://doi.org/10.18653/v1/2025.findings-acl.823
%P 15974-16007
Markdown (Informal)
[Token-level Preference Self-Alignment Optimization for Multi-style Outline Controllable Generation](https://aclanthology.org/2025.findings-acl.823/) (Li et al., Findings 2025)
ACL