Model scheduling processing method and device, equipment and storage medium

Abstract

The invention provides a model scheduling processing method and device, equipment and a storage medium, and the method comprises the steps: determining the performance parameters of each network model according to the operation log data of each network model during the execution of a single-path serial task and the operation log data of each network model during the execution of a multi-path parallel task; and determining a scheduling strategy for each network model according to the performance parameters of each network model and the operation log data when each network model executes the target task. According to the scheme, the scheduling strategy of each network model is determined by analyzing the performance parameters of the network models and the operation log data of each network model during execution of the target task, analysis and adjustment of model scheduling are realized, and improvement of the performance of cooperative work application of multiple network models is facilitated.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, and in particular to a model scheduling processing method, device, equipment and storage medium. Background technique [0002] With the continuous improvement of hardware computing power and the complexity and diversification of application scenarios, the structure and parameter quantity of neural network models increase accordingly, and the application scenarios of multi-network models working together emerge in endlessly. For example, in the image-text translation application scenario, the task process involved includes OCR image-text recognition and text translation, and OCR image-text recognition may involve image cutting, image regularization, image recognition, etc. The neural network models used in the entire business include: Several, there are scheduling dependencies in timing between these several models. In such an application with multiple network scheduling...

AI Technical Summary

Problems solved by technology

At present, the optimization of multi-network model applications mainly focuses on the analysis and optimization of the performance of the internal operators of the model, completely ignoring the analysis and adjustment of model scheduling, which limits the performance improvement of multi-network model collaborative work applications.

Images