[2023 CANN Training Camp Season 1] MindSpore Model Quick Tuning Guide Chapter 3 - Debugging and Tuning on MindSpore Cloud

1. Debugging and tuning on the ModelArts cloud

ModelArts key initialization

Detailed tutorial:

Initialize OBS service

Create a training job

2. MindSpore IDE plug-in efficiency improvement

Through features such as intelligent code block recommendation and automatic code completion, the development efficiency of MindSpore scripts is improved, and the ModelArts cloud service is connected to realize the end-cloud collaboration of model training and deployment.
MindSpore IDE plugin installation:

1. PyCharm Settings->Plugins->Install Plugin from Disk
2. Restart PyCharm

Features of MindSpore IDE

Code Completion
Provides MindSpore code completion capability based on user code.
TreeView
provides a visual display of the code structure, which can quickly realize code navigation and improve the efficiency of script development.
Search
Supports the search capability of keywords such as MindSpore API, code, and data sets to assist users in writing code.

MindSpore IDE - Integrated MindInsight

Integrate ModelArts cloud service

1. The IDE supports configuring the ModelArts key,
   which can directly deploy the local training code to
   the cloud for execution;
2. 
   Supports one-click start of MindSpore training jobs and
   MindInsight visualization jobs on the ModelArts cloud ;
3. Supports starting MindInsight visualization
   jobs independently.