TensorSLM: Energy-efficient Embedding Compression of Sub-billion Parameter Language Models on Low-end Devices

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xu, Mingxue, Xu, Yao Lei, Mandic, Danilo P.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912432744562688
author Xu, Mingxue
Xu, Yao Lei
Mandic, Danilo P.
author_facet Xu, Mingxue
Xu, Yao Lei
Mandic, Danilo P.
contents Small Language Models (SLMs, or on-device LMs) have significantly fewer parameters than Large Language Models (LLMs). They are typically deployed on low-end devices, like mobile phones and single-board computers. Unlike LLMs, which rely on increasing model size for better generalisation, SLMs designed for edge applications are expected to have adaptivity to the deployment environments and energy efficiency given the device battery life constraints, which are not addressed in datacenter-deployed LLMs. This paper addresses these two requirements by proposing a training-free token embedding compression approach using Tensor-Train Decomposition (TTD). Each pre-trained token embedding vector is converted into a lower-dimensional Matrix Product State (MPS). We comprehensively evaluate the extracted low-rank structures across compression ratio, language task performance, latency, and energy consumption on a typical low-end device, i.e. Raspberry Pi. Taking the sub-billion parameter versions of GPT-2/Cerebres-GPT and OPT models as examples, our approach achieves a comparable language task performance to the original model with around $2.0\times$ embedding layer compression, while the energy consumption of a single query drops by half.
format Preprint
id arxiv_https___arxiv_org_abs_2506_13514
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TensorSLM: Energy-efficient Embedding Compression of Sub-billion Parameter Language Models on Low-end Devices
Xu, Mingxue
Xu, Yao Lei
Mandic, Danilo P.
Computation and Language
Machine Learning
Numerical Analysis
Small Language Models (SLMs, or on-device LMs) have significantly fewer parameters than Large Language Models (LLMs). They are typically deployed on low-end devices, like mobile phones and single-board computers. Unlike LLMs, which rely on increasing model size for better generalisation, SLMs designed for edge applications are expected to have adaptivity to the deployment environments and energy efficiency given the device battery life constraints, which are not addressed in datacenter-deployed LLMs. This paper addresses these two requirements by proposing a training-free token embedding compression approach using Tensor-Train Decomposition (TTD). Each pre-trained token embedding vector is converted into a lower-dimensional Matrix Product State (MPS). We comprehensively evaluate the extracted low-rank structures across compression ratio, language task performance, latency, and energy consumption on a typical low-end device, i.e. Raspberry Pi. Taking the sub-billion parameter versions of GPT-2/Cerebres-GPT and OPT models as examples, our approach achieves a comparable language task performance to the original model with around $2.0\times$ embedding layer compression, while the energy consumption of a single query drops by half.
title TensorSLM: Energy-efficient Embedding Compression of Sub-billion Parameter Language Models on Low-end Devices
topic Computation and Language
Machine Learning
Numerical Analysis
url https://arxiv.org/abs/2506.13514