A data processing method, an electronic device, and a storage medium

Abstract

The application relates to the technical field of artificial intelligence, and discloses a data processing method, an electronic device and a storage medium. In the method, part of cascaded nodes in a model is divided into a node set, parameters corresponding to all nodes in the node set are divided into a parameter set, a plurality of parameter sets are obtained, and / or input data of any node in the model is divided into a plurality of data sets. When a multi-core NPU performs data processing, a plurality of cores in the multi-core NPU can be used to simultaneously load different parameter sets corresponding to different node sets, and the input data of the node set can be processed respectively, and / or a plurality of cores in the multi-core NPU can be used to simultaneously load parameter sets corresponding to the same node, and different data sets of the node can be processed respectively. In this way, the waste of computing power resources of the multi-core NPU can be reduced under the condition that the input and output data before and after the model is divided are the same.

Images