Training log visualization system, method and device for deep learning training platform

Abstract

The invention discloses a training log visualization system for a deep learning training platform, and the system comprises a tool selection module used for selecting a preset visualization tool corresponding to a target training task when a first visualization request for the target training task is received; a tool loading module used for creating a target resource, loading the selected visualization tool on the target resource and exposing a positioning address of the target resource; and a visualization display module used for starting the visualization tool through the positioning addresswhen a second visual request for the target training task is received, and displaying a training log of the target training task. Through the scheme of the invention, the visualization of a traininglog is realized. The operation is simple and convenient, and the working efficiency of users is improved. The invention further discloses a training log visualization method and device of the deep learning training platform, and the corresponding technical effects are achieved.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a training log visualization system, method and device for a deep learning training platform. Background technique [0002] Container cluster is a technology that abstracts multiple physical machines into a single logical container. It is a virtualization technology that saves computing resources and is more flexible than virtual machine technology. With the rapid development of Internet applications, it is becoming more and more common to implement user applications through container clusters. Kubernetes is a commonly used open source container cluster management system, which is used to manage containerized applications on multiple hosts in the cloud platform, and can provide containerized applications with a series of complete functions such as deployment and operation, resource scheduling, service discovery, and dynamic scaling. . [0003] Deep learning algorithm ...

AI Technical Summary

Problems solved by technology

However, deep learning algorithm personnel can only obtain the training model after training, and in practical applications, some algorithm personnel may also need log data generated during the training process to analyze and improve the training model

However, in the current cloud platform, deep learning algorithm personnel can only obtain the trained training model, so it cannot meet the requirement of some algorithm personnel to obtain training logs

Images