paper-with-me

Papers

Differentiable Dynamic Quantization with Mixed Precision and Adaptive Resolution

2021-06-04 · Zhang Zhaoyang, Shao Wenqi, Gu Jinwei, Wang Xiaogang, Luo Ping

Model quantization is challenging due to many tedious hyper-parameters such as precision (bitwidth), dynamic range (minimum and maximum discrete values) and stepsize (interval between discrete values). Unlike prior arts that carefully tune these values, we present a fully differentiable approach to learn all of them, named Differentiable Dynamic Quantization (DDQ), which has several benefits. (1) DDQ is able to quantize challenging lightweight architectures like MobileNets, where different layers prefer different quantization parameters. (2) DDQ is hardware-friendly and can be easily implemented using low-precision matrix-vector multiplication, making it capable in many hardware such as ARM. (3) Extensive experiments show that DDQ outperforms prior arts on many networks and benchmarks, especially when models are already efficient and compact. e.g., DDQ is the first approach that achieves lossless 4-bit quantization for MobileNetV2 on ImageNet.

📄 PDF Abstract BibTeX arXiv:2106.02295

Code (0)

등록된 구현이 없습니다.

Tasks

Quantization

Methods 이 논문이 사용한 방법론

Batch Normalization 설명 없음
Convolution A convolution is a type of matrix operation, consisting of a kernel, a small matrix of weights, that slides over input data performing element-wise multiplication with the…
Depthwise Convolution Depthwise Convolution is a type of convolution where we apply a single convolutional filter for each input channel. In the regular 2D…
Average Pooling 설명 없음
1x1 Convolution A 1 x 1 Convolution is a convolution with some special properties in that it can be used for dimensionality reduction,…
Pointwise Convolution Pointwise Convolution is a type of convolution that uses a 1x1 kernel: a kernel that iterates through every single point. This…
Depthwise Separable Convolution While standard convolution performs the channelwise and spatial-wise computation in one step, Depthwise Separable Convolution
Inverted Residual Block 설명 없음

Similar Papers 제목 키워드 기반

CSQ: Growing Mixed-Precision Quantization Scheme with Bi-level Continuous Sparsification

2022-12-06 · Lirui Xiao, Huanrui Yang, Zhen Dong, Kurt Keutzer 외

Mixed-precision quantization has been widely applied on deep neural networks (DNNs) as it leads to significantly better efficiency-accuracy tradeoffs compared to uniform quantization. Meanwhile, determining the exact pre…

Quantization

BSQ: Exploring Bit-Level Sparsity for Mixed-Precision Neural Network Quantization

2021-02-20 · ICLR 2021 1 · Huanrui Yang, Lin Duan, Yiran Chen, Hai Li

Mixed-precision quantization can potentially achieve the optimal tradeoff between performance and compression rate of deep neural networks, and thus, have been widely investigated. However, it lacks a systematic method t…

Neural Architecture SearchQuantization

SDQ: Stochastic Differentiable Quantization with Mixed Precision

2022-06-09 · Xijie Huang, Zhiqiang Shen, Shichao Li, Zechun Liu 외

In order to deploy deep models in a computationally efficient manner, model quantization approaches have been frequently used. In addition, as new hardware that supports mixed bitwidth arithmetic operations, recent resea…

Knowledge DistillationNeural Architecture SearchQuantization

Data Quality-aware Mixed-precision Quantization via Hybrid Reinforcement Learning

2023-02-09 · Yingchun Wang, Jingcai Guo, Song Guo, Weizhan Zhang

Mixed-precision quantization mostly predetermines the model bit-width settings before actual training due to the non-differential bit-width sampling process, obtaining sub-optimal performance. Worse still, the convention…

Quantizationreinforcement-learningReinforcement LearningReinforcement Learning (RL)

Scale When Needed: Adaptive Neuron-level Mixed Precision Quantization Aware Training

2026-05-24 · Ayush K. Varshney, Konstantinos Vandikas, Šarūnas Girdzijauskas, Adam Orucu 외 arxiv

Deploying deep neural networks on resource-constrained 6G edge devices demands aggressive compression with minimal accuracy loss. Quantization-Aware Training (QAT) has emerged as a leading compression approach; however, …