Creating a Visualization Job¶
Sample Code¶
In the ModelArts notebook instance, you do not need to enter authentication parameters for session authentication. For details about session authentication of other development environments, see Session Authentication.
from modelarts.session import Session
from modelarts.estimator import VisualizationJob
session = Session()
job = VisualizationJob(modelarts_session=session)
job_visualization_instance = job.create_visualization_job(train_url='/bucket/train/', job_name='visualization_job', job_desc='my visualization job')
Parameter Description¶
Parameter | Mandatory | Type | Description |
|---|---|---|---|
job_name | No | String | Name of a visualization job. The value is a string of 1 to 20 characters consisting of only digits, letters, underscores (_), and hyphens (-). |
job_desc | No | String | Description of a visualization job. The value is a string of 0 to 256 characters. By default, this parameter is left blank. |
train_url | Yes | String | OBS path to the visualization file. The visualization file is provided for the visualization job to read and display, and is usually located in the training output path. The visualization file is generated by the tf.summary or tensorboardx.SummaryWriter module in the training code, and the file name usually starts with events.out.tfevents. |
Parameter | Type | Description |
|---|---|---|
VisualizationJob | Object | Visualization job object. This object contains attributes such as visualization_id, create_time, job_name, and status, and operations on a visualization job, such as querying, modifying, stopping, restarting, or deleting the visualization job. |
Parameter | Type | Description |
|---|---|---|
create_time | Long | Time when a visualization job is created |
job_name | String | Name of a visualization job |
status | Byte | Status of a visualization job. For details about the job statuses, see Job Statuses. |
job_id | String | ID of a visualization job |
is_success | Boolean | Whether the API call succeeds |