All Products
Search
Document Center

Platform For AI:Register and manage models

Last Updated:Sep 13, 2024

You can register a trained model as a new model and upgrade the version of the model on the Model Management page. In this way, you can manage models in the Machine Learning Platform for AI (PAI) console in a centralized manner. This topic describes how to register and manage a model.

Prerequisites

A model is trained and uploaded to an Object Storage Service (OSS) bucket. You can train and export a model with Machine Learning Designer or Data Science Workshop (DSW). After that, you can upload the model to OSS, and then register and manage the model. For more information, see Upload objects.

Prepare an operation account

  • Alibaba Cloud account: You can use an Alibaba Cloud account to complete all operations without additional authorization.

  • RAM user: You need to add a RAM user as a workspace member of certain roles and assign permissions to the roles. For more information, see Appendix: Roles and permissions.

Register a model

You can register a new model or upgrade the version of an existing model.

  1. Go to the Model Management page.

    1. Log on to the PAI console.

    2. In the left-side navigation pane, click Workspaces. On the Workspaces page, click the name of the workspace that you want to manage.

    3. In the left-side navigation pane, choose AI Computing Asset Management Models.

  2. On the Model page, click Register Model .

  3. In the New model registration panel, configure the parameters and click Determine.

    The following registration methods are available:

    • New model: registers a trained model as a new model.

    • New version: updates the version of a registered model.

    The following table describes the key parameters.

    Parameter

    Description

    Model name

    This parameter is available only if you select New model.

    Select Model

    This parameter is available only if you select New version.

    Select a registered model.

    Version number

    Enter the version number of the registered model. This parameter is available only if you select New version.

    Model address

    You can use one of the following methods to configure the model address:

    • Enter the OSS path of the model file in the input box.

    • Click 文件夹 and select the OSS path that stores the model file.

      In the Select OSS directory or file dialog box, you can select an existing model file or perform the following steps to upload a local model file.

      1. Optional. If you want to upload the model file to a new directory, click Create Directory to create a directory in the current OSS bucket. Double-click the directory to go to the directory.

      2. Click Upload file.

      3. Click View local files and select the model file that you want to upload, or directly drag and drop the model file to the blank area.

    Model Service Deployment

    You can set this parameter in one of the following ways:

    • Automatic Configurations: After you configure Model format, the system automatically generates a deployment configuration.

    • Custom Configuration: You can enter the deployment configuration in the field. For more information, see Model deployment and deployment configurations.

    Advanced Settings

    Click Advanced Settings and configure the following parameters:

    • Version Approval Status

      Valid values:

      • Pending (default)

      • Approved

    • Model Metrics: You can configure model metrics to compare model versions. For more information, see Model metrics.

Manage a model

On the Model Management page, you can view all models that you have permissions, and view or delete a model. 模型管理

  • You can find the model that you want to manage and click View in the Operation column to go to the OSS path of this model and view the model details. You can also click Delete to delete the model.

  • You can add labels to models and then filter models by label key or label value.

  • You can click the setting icon in the upper-right corner of the page to specify the columns that you want to display in the model list.

  • You can move the pointer over the icon displayed in the ④ section as shown in the following figure to view the model ID.