A multi-computing model management method for complex scenarios

Abstract

The invention discloses a multi-computing model management method oriented to complex scenarios, relates to the technical field of model management, and is used for solving the problems of poor generality, low model computing throughput rate and wasting computing resources in the multi-computing model management method in the prior art, The present invention includes creating a process calculation graph, which is used to show the topological relationship between each process node, the flow direction of data and messages in the model calculation process; data reception, by the communication process in the process calculation graph Receive external data, and process the data and pass it to the model process; for model calculation, the model process uses the received data to calculate the result and continues to pass it to the lower node until it is sent back to the communication process; The result is sent to an external system. In the present invention, through the above technical solutions, the throughput rate of model calculation can be improved, the allocation and management of computing resources can be completed, and the computing resources can be greatly saved.

Description

technical field [0001] The invention relates to the technical field of management of computing models, and more particularly to a method for managing multiple computing models oriented to complex scenarios. Background technique [0002] With the application and development of deep learning in various fields, a variety of excellent artificial intelligence products continue to emerge, and the most important of these products is the deep learning model, and people are concerned about the speed, efficiency and performance of the deep learning model. The requirements are getting higher and higher, and the application scenarios of deep learning models are becoming more and more complex, so it is necessary to manage the deep learning models of complex scenarios accordingly. [0003] Existing deep learning model management methods are mainly to package all trained computing models before deployment, fix the order of execution and dependencies between the computing models in the code...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is: in order to solve the problems of poor versatility, low model calculation throughput, and waste of computing resources in the multi-computing model management method in the prior art, the present invention provides a multi-computing model management method for complex scenarios to improve The versatility of multi-calculation model management methods improves the throughput of model calculations and saves computing resources

Images