Hardware-Centric AutoML for Mixed-Precision Quantization

Abstract

Model quantization is a widely used technique to compress and accelerate deep neural network (DNN) inference. Emergent DNN hardware accelerators begin to support mixed precision (1-8 bits) to further improve the computation efficiency, which raises a great challenge to find the optimal bitwidth for each layer: it requires domain experts to explore the vast design space trading off among accuracy, latency, energy, and model size, which is both time-consuming and sub-optimal. Conventional quantization algorithm ignores the different hardware architectures and quantizes all the layers in a uniform way. In this paper, we introduce the Hardware-Aware Automated Quantization (HAQ) framework which leverages the reinforcement learning to automatically determine the quantization policy, and we take the hardware accelerator's feedback in the design loop. Rather than relying on proxy signals such as FLOPs and model size, we employ a hardware simulator to generate direct feedback signals (latency and energy) to the RL agent. Compared with conventional methods, our framework is fully automated and can specialize the quantization policy for different neural network architectures and hardware architectures. Our framework effectively reduced the latency by 1.4-1.95x and the energy consumption by 1.9x with negligible loss of accuracy compared with the fixed bitwidth (8 bits) quantization. Our framework reveals that the optimal policies on different hardware architectures (i.e., edge and cloud architectures) under different resource constraints (i.e., latency, energy, and model size) are drastically different. We interpreted the implication of different quantization policies, which offer insights for both neural network architecture design and hardware architecture design.

Journal preprint of arXiv:1811.08886 (IJCV, 2020). The first three authors contributed equally to this work. Project page: https://hanlab.mit.edu/projects/haq/

References in corpus (16)

• Continuous control with deep reinforcement learning
• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• Neural Architecture Search with Reinforcement Learning
• Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding
• Binarized Neural Networks: Training Deep Neural Networks with Weights and Activations Constrained to +1 or -1
• DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients
• AMC: AutoML for Model Compression and Acceleration on Mobile Devices
• Quantizing deep convolutional networks for efficient inference: A whitepaper
• Trained Ternary Quantization
• PACT: Parameterized Clipping Activation for Quantized Neural Networks
• Efficient Neural Architecture Search via Parameter Sharing
• Channel Pruning for Accelerating Very Deep Neural Networks
• ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware
• Learning Efficient Convolutional Networks through Network Slimming
• Path-Level Network Transformation for Efficient Architecture Search
• NetAdapt: Platform-Aware Neural Network Adaptation for Mobile Applications

Cited by in corpus (3)

• Enable Deep Learning on Mobile Devices: Methods, Systems, and Applications
• PVNAS: 3D Neural Architecture Search with Point-Voxel Convolution
• FxP-QNet: A Post-Training Quantizer for the Design of Mixed Low-Precision DNNs with Dynamic Fixed-Point Representation