Data processing method, model optimization device and model execution device

Abstract

The embodiment of the invention discloses a data processing method in the field of artificial intelligence, which is used for reducing the loading duration of a neural network model. The method provided by the embodiment of the invention comprises the steps of obtaining a neural network model; determining a memory size of a memory space required for reasoning operation based on the neural networkmodel; and updating the neural network model to obtain a target neural network model, the target neural network model carrying information indicating the memory size.

Description

technical field [0001] The present application relates to the field of artificial intelligence, in particular to a data processing method, a model optimization device and a model execution device. Background technique [0002] Deep Learning (DL) is an important technology in the field of Artificial Intelligence (AI). DL can usually be divided into two processes: Training and Inference. In the training process, a large amount of sample data is input, and a complex deep neural network model is trained by using unsupervised learning methods such as reinforcement learning. The reasoning process uses the trained model and new data to "reason" various conclusions. For example, the video surveillance equipment judges whether a captured face belongs to the blacklist through the trained deep neural network model. Due to the huge amount of training data and complex deep neural network structure involved in the training process, the calculation is huge, and it is usually completed by...

AI Technical Summary

Problems solved by technology

[0004] It can be seen that the existing technology pre-calculates the memory required for the operation of the neural network model based on the memory allocation and multiplexing algorithms deployed on the smart device, thereby achieving the purpose of reducing the number of memory allocation and release operations. However, when the resources of the smart device are limited , it takes a lot of precomputation time, causing the model to load too long

Images