Qualcomm Incorporated

Andrey Kuzmin

Tijmen Pieter Frederik Blankevoort

Markus Nagel

Marinus Willem Van Baalen

Andrii Skliar

Brian Kahne

A processor-implemented method includes bit shifting a binary representation of a neural network parameter. The neural network parameter has fewer bits, b, than a number of hardware bits, B, supported by hardware that processes the neural network parameter. The bit shifting effectively multiplies the neural network parameter by 2. The method also includes dividing a quantization scale by 2to obtain an updated quantization scale. The method further includes quantizing the bit shifted binary representation with the updated quantization scale to obtain a value for the neural network parameter.

Growing community of inventors

Amsterdam, Netherlands

Andrey Kuzmin