AdaBits: Neural Network Quantization with Adaptive Bit-Widths
Preprint
Abstract
Deep neural networks with adaptive configurations have gained increasing attention due to the instant and flexible deployment of these models on platforms with different resource budgets. In this paper, we investigate a novel option to achieve this goal by enabling adaptive bit-widths of weights and activations in the model. We first examine the benefits and challenges of training quantized model with adaptive bit-widths, and then experiment with several approaches including direct adaptation, progressive training and joint training. We discover that joint training is able to produce comparable performance on the adaptive model as individual models. We further propose a new technique named Switchable Clipping Level (S-CL) to further improve quantized models at the lowest bit-width. With our proposed techniques applied on a bunch of models including MobileNet-V1/V2 and ResNet-50, we demonstrate that bit-width of weights and activations is a new option for adaptively executable deep neural networks, offering a distinct opportunity for improved accuracy-efficiency trade-off as well as instant adaptation according to the platform constraints in real-world applications.Keywords
All Related Versions
This publication has 5 references indexed in Scilit:
- Single-Path NAS: Designing Hardware-Efficient ConvNets in Less Than 4 Hours2020
- A survey of deep learning techniques for autonomous drivingJournal of Field Robotics, 2019
- AutoGAN: Neural Architecture Search for Generative Adversarial Networks2019
- Volumetric Medical Image Segmentation: A 3D Deep Coarse-to-Fine Framework and Its Adversarial Examples2019
- Balanced Quantization: An Effective and Efficient Approach to Quantized Neural NetworksJournal of Computer Science and Technology, 2017