@inproceedings{qin-etal-2025-federated,
title = "Federated Data-Efficient Instruction Tuning for Large Language Models",
author = "Qin, Zhen and
Wu, Zhaomin and
He, Bingsheng and
Deng, Shuiguang",
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.803/",
doi = "10.18653/v1/2025.findings-acl.803",
pages = "15550--15568",
ISBN = "979-8-89176-256-5",
abstract = "Instruction tuning is a crucial step in improving the responsiveness of pretrained large language models (LLMs) to human instructions. Federated learning (FL) helps to exploit the use of vast private instruction data from clients, becoming popular for LLM tuning by improving data diversity. Existing federated tuning simply consumes all local data, causing excessive computational overhead and overfitting to local data, while centralized data-efficient solutions are not suitable for FL due to privacy concerns. This work presents FedHDS, a federated data-efficient instruction tuning approach, which tunes LLMs with a representative subset of edge-side data. It reduces the data redundancy at both intra- and inter-client levels without sharing raw data. Experiments with various LLMs, datasets and partitions show that FedHDS improves Rouge-L on unseen tasks by an average of 10.72{\%} over the SOTA full-data federated instruction tuning methods, while using less than 1.5{\%} of the data samples, improving training efficiency by up to tens of times."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="qin-etal-2025-federated">
<titleInfo>
<title>Federated Data-Efficient Instruction Tuning for Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Qin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhaomin</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bingsheng</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shuiguang</namePart>
<namePart type="family">Deng</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>Instruction tuning is a crucial step in improving the responsiveness of pretrained large language models (LLMs) to human instructions. Federated learning (FL) helps to exploit the use of vast private instruction data from clients, becoming popular for LLM tuning by improving data diversity. Existing federated tuning simply consumes all local data, causing excessive computational overhead and overfitting to local data, while centralized data-efficient solutions are not suitable for FL due to privacy concerns. This work presents FedHDS, a federated data-efficient instruction tuning approach, which tunes LLMs with a representative subset of edge-side data. It reduces the data redundancy at both intra- and inter-client levels without sharing raw data. Experiments with various LLMs, datasets and partitions show that FedHDS improves Rouge-L on unseen tasks by an average of 10.72% over the SOTA full-data federated instruction tuning methods, while using less than 1.5% of the data samples, improving training efficiency by up to tens of times.</abstract>
<identifier type="citekey">qin-etal-2025-federated</identifier>
<identifier type="doi">10.18653/v1/2025.findings-acl.803</identifier>
<location>
<url>https://aclanthology.org/2025.findings-acl.803/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>15550</start>
<end>15568</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Federated Data-Efficient Instruction Tuning for Large Language Models
%A Qin, Zhen
%A Wu, Zhaomin
%A He, Bingsheng
%A Deng, Shuiguang
%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 qin-etal-2025-federated
%X Instruction tuning is a crucial step in improving the responsiveness of pretrained large language models (LLMs) to human instructions. Federated learning (FL) helps to exploit the use of vast private instruction data from clients, becoming popular for LLM tuning by improving data diversity. Existing federated tuning simply consumes all local data, causing excessive computational overhead and overfitting to local data, while centralized data-efficient solutions are not suitable for FL due to privacy concerns. This work presents FedHDS, a federated data-efficient instruction tuning approach, which tunes LLMs with a representative subset of edge-side data. It reduces the data redundancy at both intra- and inter-client levels without sharing raw data. Experiments with various LLMs, datasets and partitions show that FedHDS improves Rouge-L on unseen tasks by an average of 10.72% over the SOTA full-data federated instruction tuning methods, while using less than 1.5% of the data samples, improving training efficiency by up to tens of times.
%R 10.18653/v1/2025.findings-acl.803
%U https://aclanthology.org/2025.findings-acl.803/
%U https://doi.org/10.18653/v1/2025.findings-acl.803
%P 15550-15568
Markdown (Informal)
[Federated Data-Efficient Instruction Tuning for Large Language Models](https://aclanthology.org/2025.findings-acl.803/) (Qin et al., Findings 2025)
ACL