JZUS - Journal of Zhejiang University SCIENCE

Frontiers of Information Technology & Electronic Engineering

Accepted manuscript available online (unedited version)

Adaptive layer splitting for wireless LLM inference in edge computing: a model-based reinforcement learning approach

Author(s): Yuxuan CHEN, Rongpeng LI, Xiaoxue YU, Zhifeng ZHAO, Honggang ZHANG
Affiliation(s): College of Information Science & more
Corresponding email(s): cyx00@zju.edu.cn, lirongpeng@zju.edu.cn, sdwhyxx@zju.edu.cn, zhaozf@zhejianglab.com, honggangzhang@zju.edu.cn
Key Words: Large language models (LLMs); Edge computing; Model-based reinforcement learning (MBRL); Split inference; Transformer

Share this article to： More <<< Previous Paper \|Next Paper >>>

Yuxuan CHEN, Rongpeng LI, Xiaoxue YU, Zhifeng ZHAO, Honggang ZHANG. Adaptive layer splitting for wireless LLM inference in edge computing: a model-based reinforcement learning approach[J]. Frontiers of Information Technology & Electronic Engineering,in press.https://doi.org/10.1631/FITEE.2400468

@article{title="Adaptive layer splitting for wireless LLM inference in edge computing: a model-based reinforcement learning approach",
author="Yuxuan CHEN, Rongpeng LI, Xiaoxue YU, Zhifeng ZHAO, Honggang ZHANG",
journal="Frontiers of Information Technology & Electronic Engineering",
year="in press",
publisher="Zhejiang University Press & Springer",
doi="https://doi.org/10.1631/FITEE.2400468"
}

%0 Journal Article
%T Adaptive layer splitting for wireless LLM inference in edge computing: a model-based reinforcement learning approach
%A Yuxuan CHEN
%A Rongpeng LI
%A Xiaoxue YU
%A Zhifeng ZHAO
%A Honggang ZHANG
%J Frontiers of Information Technology & Electronic Engineering
%P
%@ 2095-9184
%D in press
%I Zhejiang University Press & Springer
doi="https://doi.org/10.1631/FITEE.2400468"

TY - JOUR
T1 - Adaptive layer splitting for wireless LLM inference in edge computing: a model-based reinforcement learning approach
A1 - Yuxuan CHEN
A1 - Rongpeng LI
A1 - Xiaoxue YU
A1 - Zhifeng ZHAO
A1 - Honggang ZHANG
J0 - Frontiers of Information Technology & Electronic Engineering
SP -
EP -
%@ 2095-9184
Y1 - in press
PB - Zhejiang University Press & Springer
ER -
doi="https://doi.org/10.1631/FITEE.2400468"

Abstract
Chinese Summary
Academic Network
Reviewer Comment

Abstract: Optimizing the deployment of large language models (LLMs) in edge computing environments is critical for enhancing privacy and computational efficiency. In the path toward efficient wireless LLM inference in edge computing, this study comprehensively analyzes the impact of different splitting points in mainstream open-source LLMs. Accordingly, this study introduces a framework taking inspiration from model-based reinforcement learning (MBRL) to determine the optimal splitting point across the edge and user equipment (UE). By incorporating a reward surrogate model, our approach significantly reduces the computational cost of frequent performance evaluations. Extensive simulations demonstrate that this method effectively balances inference performance and computational load under varying network conditions, providing a robust solution for LLM deployment in decentralized settings.

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference

Open peer comments: Debate/Discuss/Question/Opinion

<1>

- Go to

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference