HAWQV3: Dyadic Neural Network Quantization

Abstract

Current low-precision quantization algorithms often have the hidden cost of conversion back and forth from floating point to quantized integer values. This hidden cost limits the latency improvement realized by quantizing Neural Networks. To address this, we present HAWQV3, a novel mixed-precision integer-only quantization framework. The contributions of HAWQV3 are the following: (i) An integer-only inference where the entire computational graph is performed only with integer multiplication, addition, and bit shifting, without any floating point operations or even integer division; (ii) A novel hardware-aware mixed-precision quantization method where the bit-precision is calculated by solving an integer linear programming problem that balances the trade-off between model perturbation and other constraints, e.g., memory footprint and latency; (iii) Direct hardware deployment and open source contribution for 4-bit uniform/mixed-precision quantization in TVM, achieving an average speed up of $1.45\times$ for uniform 4-bit, as compared to uniform 8-bit for ResNet50 on T4 GPUs; and (iv) extensive evaluation of the proposed methods on ResNet18/50 and InceptionV3, for various model compression levels with/without mixed precision. For ResNet50, our INT8 quantization achieves an accuracy of $77.58\%$, which is $2.68\%$ higher than prior integer-only work, and our mixed-precision INT4/8 quantization can reduce INT8 latency by $23\%$ and still achieve $76.73\%$ accuracy. Our framework and the TVM implementation have been open sourced.

References in corpus (23)

• Distilling the Knowledge in a Neural Network
• EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
• MXNet: A Flexible and Efficient Machine Learning Library for Heterogeneous Distributed Systems
• DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradients
• Pruning Convolutional Neural Networks for Resource Efficient Inference
• cuDNN: Efficient Primitives for Deep Learning
• Binarized Neural Networks
• Quantizing deep convolutional networks for efficient inference: A whitepaper
• PACT: Parameterized Clipping Activation for Quantized Neural Networks
• Pruning Filters for Efficient ConvNets
• Model compression via distillation and quantization
• Tensor Comprehensions: Framework-Agnostic High-Performance Machine Learning Abstractions
• Exploring the Regularity of Sparse Structure in Convolutional Neural Networks
• Mixed Precision Quantization of ConvNets via Differentiable Neural Architecture Search
• Apprentice: Using Knowledge Distillation Techniques To Improve Low-Precision Network Accuracy
• TVM: An Automated End-to-End Optimizing Compiler for Deep Learning
• Bayesian Bits: Unifying Quantization and Pruning
• Improving Post Training Neural Quantization: Layer-wise Calibration and Integer Programming
• PyHessian: Neural Networks Through the Lens of the Hessian
• Efficient Execution of Quantized Deep Learning Models: A Compiler Approach
• Adaptive Quantization for Deep Neural Network
• On Periodic Functions as Regularizers for Quantization of Neural Networks
• One Weight Bitwidth to Rule Them All

Cited by in corpus (7)

• Applications and Techniques for Fast Machine Learning in Science
• I-BERT: Integer-only BERT Quantization
• Sharpness-aware Quantization for Deep Neural Networks
• OMPQ: Orthogonal Mixed Precision Quantization
• Integer-Only Neural Network Quantization Scheme Based on Shift-Batch-Normalization
• HAO: Hardware-aware neural Architecture Optimization for Efficient Inference
• HPTQ: Hardware-Friendly Post Training Quantization