@inproceedings{yao-etal-2026-edgeformer,
title = "{E}dge{F}ormer: Latency-Aware Collaborative Multi-Head Attention of Transformer Inference in Edge Networks",
author = "Yao, Yiming and
Niu, Jianwei and
Dai, Bin and
Ren, Tao",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.acl-long.2007/",
pages = "43346--43361",
ISBN = "979-8-89176-390-6",
abstract = "Recent breakthroughs in Transformer-based large models, have driven widespread tasks, yet their reliance on centralized cloud deployment raises significant privacy risks due to sensitive data exposure. While edge-based collaborative inference offers a privacy-preserving alternative, existing methods face critical limitations: static model partitioning cannot adapt to dynamic edge resource fluctuations, and rigid multi-head attention handling overlooks semantic-critical prioritization and parallelism. We propose EdgeFormer, a latency-aware framework for distributed Transformer inference in resource-constrained edge networks. EdgeFormer dynamically allocates model blocks across devices via efficiency-storage trade-off optimization and introduces collaborative Multi-Head Attention (cMHA), which distributes semantic-critical attention heads across devices while pruning redundant ones under real-time constraints. We further develop LiScore, a composite metric integrating attention diversity and latency costs, alongside a similarity-based retrieval method to reduce recomputation overhead. Extensive experiments demonstrate that EdgeFormer achieves up to 2.01 $\\times$ inference acceleration over state-of-the-art baselines with $\\leq$1.06{\%} accuracy loss, maintaining robustness under varying edge conditions."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yao-etal-2026-edgeformer">
<titleInfo>
<title>EdgeFormer: Latency-Aware Collaborative Multi-Head Attention of Transformer Inference in Edge Networks</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yiming</namePart>
<namePart type="family">Yao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianwei</namePart>
<namePart type="family">Niu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bin</namePart>
<namePart type="family">Dai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tao</namePart>
<namePart type="family">Ren</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-390-6</identifier>
</relatedItem>
<abstract>Recent breakthroughs in Transformer-based large models, have driven widespread tasks, yet their reliance on centralized cloud deployment raises significant privacy risks due to sensitive data exposure. While edge-based collaborative inference offers a privacy-preserving alternative, existing methods face critical limitations: static model partitioning cannot adapt to dynamic edge resource fluctuations, and rigid multi-head attention handling overlooks semantic-critical prioritization and parallelism. We propose EdgeFormer, a latency-aware framework for distributed Transformer inference in resource-constrained edge networks. EdgeFormer dynamically allocates model blocks across devices via efficiency-storage trade-off optimization and introduces collaborative Multi-Head Attention (cMHA), which distributes semantic-critical attention heads across devices while pruning redundant ones under real-time constraints. We further develop LiScore, a composite metric integrating attention diversity and latency costs, alongside a similarity-based retrieval method to reduce recomputation overhead. Extensive experiments demonstrate that EdgeFormer achieves up to 2.01 \\times inference acceleration over state-of-the-art baselines with \łeq1.06% accuracy loss, maintaining robustness under varying edge conditions.</abstract>
<identifier type="citekey">yao-etal-2026-edgeformer</identifier>
<location>
<url>https://aclanthology.org/2026.acl-long.2007/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>43346</start>
<end>43361</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T EdgeFormer: Latency-Aware Collaborative Multi-Head Attention of Transformer Inference in Edge Networks
%A Yao, Yiming
%A Niu, Jianwei
%A Dai, Bin
%A Ren, Tao
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-390-6
%F yao-etal-2026-edgeformer
%X Recent breakthroughs in Transformer-based large models, have driven widespread tasks, yet their reliance on centralized cloud deployment raises significant privacy risks due to sensitive data exposure. While edge-based collaborative inference offers a privacy-preserving alternative, existing methods face critical limitations: static model partitioning cannot adapt to dynamic edge resource fluctuations, and rigid multi-head attention handling overlooks semantic-critical prioritization and parallelism. We propose EdgeFormer, a latency-aware framework for distributed Transformer inference in resource-constrained edge networks. EdgeFormer dynamically allocates model blocks across devices via efficiency-storage trade-off optimization and introduces collaborative Multi-Head Attention (cMHA), which distributes semantic-critical attention heads across devices while pruning redundant ones under real-time constraints. We further develop LiScore, a composite metric integrating attention diversity and latency costs, alongside a similarity-based retrieval method to reduce recomputation overhead. Extensive experiments demonstrate that EdgeFormer achieves up to 2.01 \\times inference acceleration over state-of-the-art baselines with \łeq1.06% accuracy loss, maintaining robustness under varying edge conditions.
%U https://aclanthology.org/2026.acl-long.2007/
%P 43346-43361
Markdown (Informal)
[EdgeFormer: Latency-Aware Collaborative Multi-Head Attention of Transformer Inference in Edge Networks](https://aclanthology.org/2026.acl-long.2007/) (Yao et al., ACL 2026)
ACL