Saved in:
Bibliographic Details
Main Authors: Wang, Yanshu, Yang, Tong, Liang, Xiyan, Wang, Guoan, Lu, Hanning, Zhe, Xu, Li, Yaoming, Weitao, Li
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2409.11650
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914951721910272
author Wang, Yanshu
Yang, Tong
Liang, Xiyan
Wang, Guoan
Lu, Hanning
Zhe, Xu
Li, Yaoming
Weitao, Li
author_facet Wang, Yanshu
Yang, Tong
Liang, Xiyan
Wang, Guoan
Lu, Hanning
Zhe, Xu
Li, Yaoming
Weitao, Li
contents This paper provides a comprehensive overview of the principles, challenges, and methodologies associated with quantizing large-scale neural network models. As neural networks have evolved towards larger and more complex architectures to address increasingly sophisticated tasks, the computational and energy costs have escalated significantly. We explore the necessity and impact of model size growth, highlighting the performance benefits as well as the computational challenges and environmental considerations. The core focus is on model quantization as a fundamental approach to mitigate these challenges by reducing model size and improving efficiency without substantially compromising accuracy. We delve into various quantization techniques, including both post-training quantization (PTQ) and quantization-aware training (QAT), and analyze several state-of-the-art algorithms such as LLM-QAT, PEQA(L4Q), ZeroQuant, SmoothQuant, and others. Through comparative analysis, we examine how these methods address issues like outliers, importance weighting, and activation quantization, ultimately contributing to more sustainable and accessible deployment of large-scale models.
format Preprint
id arxiv_https___arxiv_org_abs_2409_11650
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Art and Science of Quantizing Large-Scale Models: A Comprehensive Overview
Wang, Yanshu
Yang, Tong
Liang, Xiyan
Wang, Guoan
Lu, Hanning
Zhe, Xu
Li, Yaoming
Weitao, Li
Machine Learning
Artificial Intelligence
This paper provides a comprehensive overview of the principles, challenges, and methodologies associated with quantizing large-scale neural network models. As neural networks have evolved towards larger and more complex architectures to address increasingly sophisticated tasks, the computational and energy costs have escalated significantly. We explore the necessity and impact of model size growth, highlighting the performance benefits as well as the computational challenges and environmental considerations. The core focus is on model quantization as a fundamental approach to mitigate these challenges by reducing model size and improving efficiency without substantially compromising accuracy. We delve into various quantization techniques, including both post-training quantization (PTQ) and quantization-aware training (QAT), and analyze several state-of-the-art algorithms such as LLM-QAT, PEQA(L4Q), ZeroQuant, SmoothQuant, and others. Through comparative analysis, we examine how these methods address issues like outliers, importance weighting, and activation quantization, ultimately contributing to more sustainable and accessible deployment of large-scale models.
title Art and Science of Quantizing Large-Scale Models: A Comprehensive Overview
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2409.11650