Optimizing Large Language Models through Quantization: A Comparative Analysis of PTQ and QAT Techniques

TL;DR

This paper compares PTQ and QAT techniques for large language model quantization, demonstrating significant size and efficiency improvements with minimal performance loss, and introduces a theoretical framework for mixed-precision strategies.

Contribution

It provides a comprehensive empirical evaluation of quantization methods for LLMs and proposes a novel theoretical approach for optimal mixed-precision quantization.

Findings

Up to 68% model size reduction with minimal performance loss

INT8 quantization reduces computational cost by 40%

INT4 quantization further reduces cost by 60% and improves hardware throughput

Abstract

This paper presents a comprehensive analysis of quantization techniques for optimizing Large Language Models (LLMs), specifically focusing on Post-Training Quantization (PTQ) and Quantization-Aware Training (QAT). Through empirical evaluation across models ranging from 10M to 1B parameters, we demonstrate that quantization can achieve up to 68% reduction in model size while maintaining performance within 6% of full-precision baselines when utilizing our proposed scaling factor {\gamma}. Our experiments show that INT8 quantization delivers a 40% reduction in computational cost and power consumption, while INT4 quantization further improves these metrics by 60%. We introduce a novel theoretical framework for mixed-precision quantization, deriving optimal bit allocation strategies based on layer sensitivity and weight variance. Hardware efficiency evaluations on edge devices reveal that our quantization approach enables up to 2.4x throughput improvement for INT8 and 3x for INT4, with 60% power reduction compared to full-precision models.