Quantization in Layer's Input is Matter

02/10/2022

∙

In this paper, we will show that the quantization in layer's input is more important than parameters' quantization for loss function. And the algorithm which is based on the layer's input quantization error is better than hessian-based mixed precision layout algorithm.

READ FULL TEXT

Quantization in Layer's Input is Matter

Mixed-Precision Inference Quantization: Radically Towards Faster inference speed, Lower Storage requirement, and Lower Loss

Towards the Limit of Network Quantization

Vector Quantization by Minimizing Kullback-Leibler Divergence

Image De-Quantization Using Generative Models as Priors

Stochastic Learning Equation using Monotone Increasing Resolution of Quantization

Channel-wise Hessian Aware trace-Weighted Quantization of Neural Networks

Pairwise Quantization

Quantization in Layer's Input is Matter

Related Research

Mixed-Precision Inference Quantization: Radically Towards Faster inference speed, Lower Storage requirement, and Lower Loss

Towards the Limit of Network Quantization

Vector Quantization by Minimizing Kullback-Leibler Divergence

Image De-Quantization Using Generative Models as Priors

Stochastic Learning Equation using Monotone Increasing Resolution of Quantization

Channel-wise Hessian Aware trace-Weighted Quantization of Neural Networks

Pairwise Quantization