Saved in:
Bibliographic Details
Main Authors: Sanyal, Arnab, Datta, Gourav, Mukherjee, Prithwish, Chinchali, Sandeep P., Orshansky, Michael
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.02380
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910184042921984
author Sanyal, Arnab
Datta, Gourav
Mukherjee, Prithwish
Chinchali, Sandeep P.
Orshansky, Michael
author_facet Sanyal, Arnab
Datta, Gourav
Mukherjee, Prithwish
Chinchali, Sandeep P.
Orshansky, Michael
contents Large Language Models (LLMs) achieve strong performance across tasks, but face storage and compute challenges on edge devices. We propose EntroLLM, a compression framework combining mixed quantization and entropy coding to reduce storage while preserving accuracy. We use a combination of unsigned and asymmetric quantization. Tensor-level quantization produces an entropy-reducing effect, increasing weight compressibility, and improving downstream Huffman encoding by $7\times$ (8-bit) and $11.3\times$ (4-bit) over state-of-the-art methods. Huffman coding further reduces memory bandwidth demands, while a parallel decoding strategy enables efficient weight retrieval with minimal latency. Experiments on edge-scale LLMs (smolLM-1.7B, phi3-mini-4k, mistral-7B) show up to $30\%$ storage savings over uint8 and $65\%$ over uint4 models, with $31.9-146.6\%$ faster inference on memory-limited devices like the NVIDIA JETSON P3450. EntroLLM requires no retraining and is compatible with existing post-training quantization pipelines, making it practical for edge LLM deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2505_02380
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle EntroLLM: Entropy Encoded Weight Compression for Efficient Large Language Model Inference on Edge Devices
Sanyal, Arnab
Datta, Gourav
Mukherjee, Prithwish
Chinchali, Sandeep P.
Orshansky, Michael
Machine Learning
Large Language Models (LLMs) achieve strong performance across tasks, but face storage and compute challenges on edge devices. We propose EntroLLM, a compression framework combining mixed quantization and entropy coding to reduce storage while preserving accuracy. We use a combination of unsigned and asymmetric quantization. Tensor-level quantization produces an entropy-reducing effect, increasing weight compressibility, and improving downstream Huffman encoding by $7\times$ (8-bit) and $11.3\times$ (4-bit) over state-of-the-art methods. Huffman coding further reduces memory bandwidth demands, while a parallel decoding strategy enables efficient weight retrieval with minimal latency. Experiments on edge-scale LLMs (smolLM-1.7B, phi3-mini-4k, mistral-7B) show up to $30\%$ storage savings over uint8 and $65\%$ over uint4 models, with $31.9-146.6\%$ faster inference on memory-limited devices like the NVIDIA JETSON P3450. EntroLLM requires no retraining and is compatible with existing post-training quantization pipelines, making it practical for edge LLM deployment.
title EntroLLM: Entropy Encoded Weight Compression for Efficient Large Language Model Inference on Edge Devices
topic Machine Learning
url https://arxiv.org/abs/2505.02380