Optimizing Large Language Models through Quantization: A Comparative Analysis of PTQ and QAT Techniques
This work addresses the computational and energy efficiency challenges for deploying LLMs on resource-constrained devices, presenting an incremental advancement in quantization methods.
This paper tackles the problem of optimizing Large Language Models (LLMs) through quantization, demonstrating that techniques like INT8 and INT4 can achieve up to 68% reduction in model size while maintaining performance within 6% of full-precision baselines, with up to 3x throughput improvement and 60% power reduction on edge devices.
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 γ. 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.