HPTQ: Hardware-Friendly Post Training Quantization

TL;DR

HPTQ introduces a post-training quantization framework that ensures neural network models are optimized for hardware efficiency by supporting uniform, symmetric, and power-of-two thresholds, enabling deployment on edge devices.

Contribution

The paper proposes a novel hardware-friendly post-training quantization method that combines existing techniques to meet specific hardware constraints.

Findings

Achieves competitive accuracy under hardware-friendly constraints

Supports a wide range of network architectures and tasks

Demonstrates effectiveness on classification, detection, segmentation, and pose estimation

Abstract

Neural network quantization enables the deployment of models on edge devices. An essential requirement for their hardware efficiency is that the quantizers are hardware-friendly: uniform, symmetric, and with power-of-two thresholds. To the best of our knowledge, current post-training quantization methods do not support all of these constraints simultaneously. In this work, we introduce a hardware-friendly post training quantization (HPTQ) framework, which addresses this problem by synergistically combining several known quantization methods. We perform a large-scale study on four tasks: classification, object detection, semantic segmentation and pose estimation over a wide variety of network architectures. Our extensive experiments show that competitive results can be obtained under hardware-friendly constraints.