@inproceedings{du-etal-2025-boost,
title = "Boost, Disentangle, and Customize: A Robust System2-to-System1 Pipeline for Code Generation",
author = "Du, Kounianhua and
Wang, Hanjing and
Liu, Jianxing and
Chen, Jizheng and
Dai, Xinyi and
Wang, Yasheng and
Tang, Ruiming and
Yu, Yong and
Wang, Jun and
Zhang, Weinan",
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.833/",
doi = "10.18653/v1/2025.findings-acl.833",
pages = "16194--16204",
ISBN = "979-8-89176-256-5",
abstract = "To address these limitations, we propose BDC, a novel framework that Boosts reasoning exploration via multi-agent collaboration, Disentangles heterogeneous data into specialized experts, and Customizes solutions through dynamic model composition. BDC integrates a Monte Carlo Tree-of-Agents algorithm, where multiple LLMs mutually verify and refine reasoning paths through reflection-guided pruning, enabling efficient exploration of high-quality solutions. To handle data diversity, we cluster problems by latent semantics, train composable LoRA experts on each cluster, and deploy an input-aware hypernetwork to dynamically merge these experts into tailored solvers. Experiments on APPS and CodeContest benchmarks demonstrate BDC{'}s superiority: it achieves up to 73.8{\%} accuracy on hard problems, outperforming state-of-the-art methods like LATS and RethinkMCTS by 9{--}15{\%}. This work lays the groundwork for advancing LLM capabilities in complex reasoning tasks, offering a novel System2-to-System1 solution."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="du-etal-2025-boost">
<titleInfo>
<title>Boost, Disentangle, and Customize: A Robust System2-to-System1 Pipeline for Code Generation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kounianhua</namePart>
<namePart type="family">Du</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hanjing</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianxing</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jizheng</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinyi</namePart>
<namePart type="family">Dai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yasheng</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruiming</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yong</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weinan</namePart>
<namePart type="family">Zhang</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>To address these limitations, we propose BDC, a novel framework that Boosts reasoning exploration via multi-agent collaboration, Disentangles heterogeneous data into specialized experts, and Customizes solutions through dynamic model composition. BDC integrates a Monte Carlo Tree-of-Agents algorithm, where multiple LLMs mutually verify and refine reasoning paths through reflection-guided pruning, enabling efficient exploration of high-quality solutions. To handle data diversity, we cluster problems by latent semantics, train composable LoRA experts on each cluster, and deploy an input-aware hypernetwork to dynamically merge these experts into tailored solvers. Experiments on APPS and CodeContest benchmarks demonstrate BDC’s superiority: it achieves up to 73.8% accuracy on hard problems, outperforming state-of-the-art methods like LATS and RethinkMCTS by 9–15%. This work lays the groundwork for advancing LLM capabilities in complex reasoning tasks, offering a novel System2-to-System1 solution.</abstract>
<identifier type="citekey">du-etal-2025-boost</identifier>
<identifier type="doi">10.18653/v1/2025.findings-acl.833</identifier>
<location>
<url>https://aclanthology.org/2025.findings-acl.833/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>16194</start>
<end>16204</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Boost, Disentangle, and Customize: A Robust System2-to-System1 Pipeline for Code Generation
%A Du, Kounianhua
%A Wang, Hanjing
%A Liu, Jianxing
%A Chen, Jizheng
%A Dai, Xinyi
%A Wang, Yasheng
%A Tang, Ruiming
%A Yu, Yong
%A Wang, Jun
%A Zhang, Weinan
%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 du-etal-2025-boost
%X To address these limitations, we propose BDC, a novel framework that Boosts reasoning exploration via multi-agent collaboration, Disentangles heterogeneous data into specialized experts, and Customizes solutions through dynamic model composition. BDC integrates a Monte Carlo Tree-of-Agents algorithm, where multiple LLMs mutually verify and refine reasoning paths through reflection-guided pruning, enabling efficient exploration of high-quality solutions. To handle data diversity, we cluster problems by latent semantics, train composable LoRA experts on each cluster, and deploy an input-aware hypernetwork to dynamically merge these experts into tailored solvers. Experiments on APPS and CodeContest benchmarks demonstrate BDC’s superiority: it achieves up to 73.8% accuracy on hard problems, outperforming state-of-the-art methods like LATS and RethinkMCTS by 9–15%. This work lays the groundwork for advancing LLM capabilities in complex reasoning tasks, offering a novel System2-to-System1 solution.
%R 10.18653/v1/2025.findings-acl.833
%U https://aclanthology.org/2025.findings-acl.833/
%U https://doi.org/10.18653/v1/2025.findings-acl.833
%P 16194-16204
Markdown (Informal)
[Boost, Disentangle, and Customize: A Robust System2-to-System1 Pipeline for Code Generation](https://aclanthology.org/2025.findings-acl.833/) (Du et al., Findings 2025)
ACL
- Kounianhua Du, Hanjing Wang, Jianxing Liu, Jizheng Chen, Xinyi Dai, Yasheng Wang, Ruiming Tang, Yong Yu, Jun Wang, and Weinan Zhang. 2025. Boost, Disentangle, and Customize: A Robust System2-to-System1 Pipeline for Code Generation. In Findings of the Association for Computational Linguistics: ACL 2025, pages 16194–16204, Vienna, Austria. Association for Computational Linguistics.