method and a device for determining the network structure of a depth learning network
The embodiment of the invention provides a method and a device for determining the network structure of a depth learning network, wherein, the method for determining the network structure of the depthlearning network comprises the following steps: obtaining the register bit width of a register in a...
Saved in:
Main Author | |
---|---|
Format | Patent |
Language | Chinese English |
Published |
26.02.2019
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | The embodiment of the invention provides a method and a device for determining the network structure of a depth learning network, wherein, the method for determining the network structure of the depthlearning network comprises the following steps: obtaining the register bit width of a register in a processor and the element bit width of an element in an input quantity for a designated network layer in the depth learning network; According to the preset relationship between the register bit width, element bit width, determining register bit width, element bit width and the specification parameters of network layer, the specification parameters of specified network layer. According to the specification parameters of the specified network layer, the network structure of the depth learning network is determined. The scheme can improve the running efficiency of the in-depth learning network after deployment.
本发明实施例提供了种深度学习网络的网络结构确定方法及装置,其中,该深度学习网络的网络结构确定方法包括:针对深度学习网络中的指定网络层,获取处理器中寄存器的寄存器位宽,以及输入量中元素的元素位宽;根据寄存器位宽、元素 |
---|---|
Bibliography: | Application Number: CN201710656729 |