@inproceedings{chu-etal-2025-towards,
title = "Towards Faithful Multi-step Reasoning through Fine-Grained Causal-aware Attribution Reasoning Distillation",
author = "Chu, Zheng and
Chen, Jingchang and
Wang, Zhongjie and
Tang, Guo and
Chen, Qianglong and
Liu, Ming and
Qin, Bing",
editor = "Rambow, Owen and
Wanner, Leo and
Apidianaki, Marianna and
Al-Khalifa, Hend and
Eugenio, Barbara Di and
Schockaert, Steven",
booktitle = "Proceedings of the 31st International Conference on Computational Linguistics",
month = jan,
year = "2025",
address = "Abu Dhabi, UAE",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.coling-main.157/",
pages = "2291--2315",
abstract = "Despite the remarkable reasoning capabilities demonstrated by large language models (LLM), the substantial computational overhead limits their practices. Some efforts have been directed toward distilling multi-step reasoning capabilities into smaller models through chain-of-thought (CoT). While CoT facilitates multi-step reasoning, the dependencies between reasoning steps are not always clearly discernible, which may lead to inconsistent reasoning. In this paper, we introduce fine-grained attribution reasoning distillation (FARD), which incorporates grounded citations to consolidate the relationships between reasoning steps. Specifically, FARD distills attribution reasoning rationales from LLMs to substitute CoT reasonings, which clarifies the dependencies among reasoning steps. Besides, we regularize the model`s attention pattern by leveraging the causal dependencies between reasoning steps, thereby enhancing the consistency of reasoning. Grounded attribution reasoning also enhances interpretability and verifiability, thereby facilitating faithful reasoning. We evaluate FARD on mathematical and general reasoning benchmarks. The experimental results indicate that FARD outperforms CoT distillation methods in mathematical reasoning, demonstrating its effectiveness. Furthermore, the small models trained with FARD have shown outstanding performance in out-of-distribution reasoning, proving strong generalization capabilities."
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chu-etal-2025-towards">
<titleInfo>
<title>Towards Faithful Multi-step Reasoning through Fine-Grained Causal-aware Attribution Reasoning Distillation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zheng</namePart>
<namePart type="family">Chu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jingchang</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhongjie</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guo</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qianglong</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ming</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bing</namePart>
<namePart type="family">Qin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-01</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 31st International Conference on Computational Linguistics</title>
</titleInfo>
<name type="personal">
<namePart type="given">Owen</namePart>
<namePart type="family">Rambow</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Leo</namePart>
<namePart type="family">Wanner</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marianna</namePart>
<namePart type="family">Apidianaki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hend</namePart>
<namePart type="family">Al-Khalifa</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Barbara</namePart>
<namePart type="given">Di</namePart>
<namePart type="family">Eugenio</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Steven</namePart>
<namePart type="family">Schockaert</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Abu Dhabi, UAE</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Despite the remarkable reasoning capabilities demonstrated by large language models (LLM), the substantial computational overhead limits their practices. Some efforts have been directed toward distilling multi-step reasoning capabilities into smaller models through chain-of-thought (CoT). While CoT facilitates multi-step reasoning, the dependencies between reasoning steps are not always clearly discernible, which may lead to inconsistent reasoning. In this paper, we introduce fine-grained attribution reasoning distillation (FARD), which incorporates grounded citations to consolidate the relationships between reasoning steps. Specifically, FARD distills attribution reasoning rationales from LLMs to substitute CoT reasonings, which clarifies the dependencies among reasoning steps. Besides, we regularize the model‘s attention pattern by leveraging the causal dependencies between reasoning steps, thereby enhancing the consistency of reasoning. Grounded attribution reasoning also enhances interpretability and verifiability, thereby facilitating faithful reasoning. We evaluate FARD on mathematical and general reasoning benchmarks. The experimental results indicate that FARD outperforms CoT distillation methods in mathematical reasoning, demonstrating its effectiveness. Furthermore, the small models trained with FARD have shown outstanding performance in out-of-distribution reasoning, proving strong generalization capabilities.</abstract>
<identifier type="citekey">chu-etal-2025-towards</identifier>
<location>
<url>https://aclanthology.org/2025.coling-main.157/</url>
</location>
<part>
<date>2025-01</date>
<extent unit="page">
<start>2291</start>
<end>2315</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Towards Faithful Multi-step Reasoning through Fine-Grained Causal-aware Attribution Reasoning Distillation
%A Chu, Zheng
%A Chen, Jingchang
%A Wang, Zhongjie
%A Tang, Guo
%A Chen, Qianglong
%A Liu, Ming
%A Qin, Bing
%Y Rambow, Owen
%Y Wanner, Leo
%Y Apidianaki, Marianna
%Y Al-Khalifa, Hend
%Y Eugenio, Barbara Di
%Y Schockaert, Steven
%S Proceedings of the 31st International Conference on Computational Linguistics
%D 2025
%8 January
%I Association for Computational Linguistics
%C Abu Dhabi, UAE
%F chu-etal-2025-towards
%X Despite the remarkable reasoning capabilities demonstrated by large language models (LLM), the substantial computational overhead limits their practices. Some efforts have been directed toward distilling multi-step reasoning capabilities into smaller models through chain-of-thought (CoT). While CoT facilitates multi-step reasoning, the dependencies between reasoning steps are not always clearly discernible, which may lead to inconsistent reasoning. In this paper, we introduce fine-grained attribution reasoning distillation (FARD), which incorporates grounded citations to consolidate the relationships between reasoning steps. Specifically, FARD distills attribution reasoning rationales from LLMs to substitute CoT reasonings, which clarifies the dependencies among reasoning steps. Besides, we regularize the model‘s attention pattern by leveraging the causal dependencies between reasoning steps, thereby enhancing the consistency of reasoning. Grounded attribution reasoning also enhances interpretability and verifiability, thereby facilitating faithful reasoning. We evaluate FARD on mathematical and general reasoning benchmarks. The experimental results indicate that FARD outperforms CoT distillation methods in mathematical reasoning, demonstrating its effectiveness. Furthermore, the small models trained with FARD have shown outstanding performance in out-of-distribution reasoning, proving strong generalization capabilities.
%U https://aclanthology.org/2025.coling-main.157/
%P 2291-2315
Markdown (Informal)
[Towards Faithful Multi-step Reasoning through Fine-Grained Causal-aware Attribution Reasoning Distillation](https://aclanthology.org/2025.coling-main.157/) (Chu et al., COLING 2025)
ACL