Updated on 2024-10-29 GMT+08:00

Rebuilding, Stopping, or Deleting a Training Job

Saving As an Algorithm

To modify the algorithm of a training job, click Save As Algorithm in the upper right corner of the training job details page.

On the Algorithms page, the algorithm parameters for the last training job are automatically set. You can modify the settings.

A subscribed algorithm cannot be saved as a new algorithm.

Rebuilding a Training Job

If you are not satisfied with a created training job, click Rebuild in the Operation column to rebuild it. The page for creating a training job is displayed. On this page, the parameter settings for the previous training job are automatically retained. You only need to modify certain parameter settings.

Stopping a Training Job

In the training job list, click Stop in the Operation column of a training job that is in creating, pending, or running state to stop the job.

After a training job is stopped, its billing stops on ModelArts.

A training job in completed, failed, terminated, or abnormal state cannot be stopped.

Deleting a Training Job

Release resources of a training job when not in use to avoid unnecessary charges.
  • On the Training Jobs page, click Delete in the Operation column. In the displayed dialog box, click OK to delete the training job.
  • Go to OBS and delete the OBS bucket and files used by the training job.

Searching for a Training Job

If you log in to ModelArts using an IAM account, all training jobs under this account are displayed in the training job list. To quickly search for a training job, use the following methods:

Method 1: Click Only my jobs. Then, only jobs created under the current IAM user account are displayed in the training job list.

Method 2: Search for jobs by name, ID, job type, status, creation time, algorithm, and resource pool.

Method 3: Click the refresh button in the upper right corner of the job list to refresh it.

Method 4: Configure the custom columns and other basic settings.

Figure 1 Searching for a training job