All Products
Document Center

Platform For AI:Pipeline overview

Last Updated:Feb 05, 2024

Machine Learning Designer allows you to build and debug models by using pipelines. To build a model, you can plan pipeline folders, create a pipeline, add different components to the pipeline, and then arrange the components based on the logic of your model.

Plan pipeline folders

Before you create pipelines, we recommend that you plan the folders for storing the pipelines based on their categories. A neat folder structure allows you to efficiently find, use, and manage pipelines. In normal cases, a pipeline generates a test model. We recommend that you store pipelines of the same category in one folder.

By default, Machine Learning Designer provides two folders for storing pipelines: My Pipelines and Pipelines Visible to Workspaces. Pipelines stored in the My Pipelines folder are visible only to you. Pipelines stored in the Pipelines Visible to Workspaces folder are visible to all members in the workspace. You can also create subfolders within the two folders based on your business requirements. This helps you categorize and manage pipelines. imageAfter you click a pipeline within a folder, you can view the basic information of the pipeline in the Basic Information section on the right side, such as the name of the pipeline, the visible range, and the pipeline preview.

Create a pipeline

Machine Learning Designer allows you to create a pipeline by using one of the following methods:

  • Create a blank pipeline

    If you want to create a fully customized pipeline based on your business requirements, you can create a blank pipeline. This way, you can build a pipeline model from scratch.

  • Create a pipeline from a preset template

    Machine Learning Designer provides a wide range of preset templates that you can use to build pipeline models. If your business scenario is similar to that of a template, we recommend that you use this template to create a pipeline. After that, you can configure your pipeline model with ease based on the settings of the template. For more information, see Overview.

  • Create a pipeline from a custom template

    You can publish a successfully-run pipeline as a custom template. Then, algorithm developers who are members of the current workspace can create a pipeline from the template.

  • Export or import a pipeline

    You can export a pipeline as a file and then import the pipeline file to create an identical pipeline in Machine Learning Designer.