【DSW Gallery】Getting Started with DSW Mirroring

Getting Started with DSW Mirroring
DSW relies on container technology to provide an out-of-the-box machine learning development environment. Each DSW instance uses a mirror to start the container. Jupyterlab, VSCode, and Terminal run in the container. After the user enters the container, use Jupyterlab to start the Kernel or use the shell on the Terminal Start the machine learning task by command to start the process. The PAI team has preset a series of official images for different operating systems, Python versions, Tensorflow or PyTorch versions, and Cuda versions to facilitate users to quickly start the service. DSW also supports using user-defined images to start services. Modifications made by users to the running environment (such as installing system libraries, using pip or conda commands to install python packages, and modifying operating system configurations) can be saved to the Aliyun ACR mirror warehouse through the save environment function.
prerequisite
• Complete DSW activation and authorization
Start an instance using an image
Enter the DSW instance creation page, you can see three Tab pages:
• Select "Official Image", and in the drop-down box, you can choose the official image preset by the PAI team to create an instance
• Select "User-defined image", and in the drop-down box, you can select a custom image uploaded by the user to be managed in the AI workspace to create an instance
• Select "Mirror URL" and fill in the public accessible Docker Registry address in the input box to create an instance
You can refer to the AIWorkspace documentation to view the currently available official images https://www.alibabacloud.com/help/zh/machine-learning-platform-for-ai/latest/image-management-1
Shutdown to save the environment
During the development process of using DSW, if you modify the environment (such as installing new tools or libraries with the apt command, and installing new python packages with the pip command), you hope that the environment modification can be saved persistently after the shutdown. Select "Stop after saving environment" when clicking the "Stop" button
save image
The running environment can also be saved to the Aliyun ACR mirror warehouse (personal version) under the user account, which can be pulled and used in other environments.
1. You can click the "Save Image" button on the console, select your ACR warehouse, fill in the name of your image Tag, and click "Save"
2. The instance status will change to "Saving"
3. After the instance is saved, the instance status changes back to "Running", and you can see the details of the instance save on the console list page