How to Parameterize Asymmetric Quantization Ranges for Quantization-Aware Training

arXiv:2404.16898v1 Announce Type: new
Abstract: This paper investigates three different parameterizations of asymmetric uniform quantization for quantization-aware training: (1) scale and offset, (2) minimum and maximum, and (3) beta and gamma. We perform a comprehensive comparative analysis of these parameterizations' influence on quantization-aware training, using both controlled experiments and real-world large language models. Our particular focus is on their changing behavior in response to critical training hyperparameters, bit width and learning rate. Based on our investigation, we propose best practices …

abstract analysis arxiv beta comparative analysis cs.ai cs.lg influence maximum minimum paper quantization scale training type uniform world