24 mars 2018 - QAT: Quantization and training of neural networks for efficient integer-arithmetic-only inference, CVPR
Description:
QAT employs the affine mapping of integers to real values with two constant parameters: Scale and Zeropoint. It first subtracts the Zero-point parameter from data (weights/activation), then divides the data by scale parameter, and finally getting the quantized results with rounding operation and affine mapping.Ajouté au bande de temps:
Date:
24 mars 2018
Maintenaint
~ Il y a 7 ans et 2 mois