The following are required to setup Vertex AI TensorBoard:

1. Create a service account with required permissions.
2. Create a Cloud Storage bucket to store Vertex AI TensorBoard logs.
3. Create a Vertex AI TensorBoard instance.

Create a service account with required permissions

The Vertex AI TensorBoard integration with custom training requiresattaching a service account.

1. Create a service account:

   gcloud--project=PROJECT_IDiamservice-accountscreateUSER_SA_NAME

   Replace the following:

   • PROJECT_ID: the ID of the project in which you'recreating a service account.

   • USER_SA_NAME: a unique name for the serviceaccount you're creating.

2. The new service account is used by the Vertex AI Training Service toaccess Google Cloud services and resources.Use the following commands to grant these roles if needed:

   Note: You need at least project level setIamPolicy permission to runthese commands. For details, seeRequired Permissions for details.

   SA_EMAIL="USER_SA_NAME@PROJECT_ID.iam.gserviceaccount.com"gcloudprojectsadd-iam-policy-bindingPROJECT_ID\--member="serviceAccount:${SA_EMAIL}"\--role="roles/storage.admin"gcloudprojectsadd-iam-policy-bindingPROJECT_ID\--member="serviceAccount:${SA_EMAIL}"\--role="roles/aiplatform.user"

Create a Cloud Storage bucket to store Vertex AI TensorBoard logs

A Cloud Storage bucket is required to store the Vertex AI TensorBoardlogs your training script generates. The bucket must beregionalthat is, not multi-region or dual-region, and the following resources mustbe in same region:

• Cloud Storage bucket
• Vertex AI training job
• Vertex AI TensorBoard instance

You can use an existing bucket instead of following the bucket creation stepdescribed here. When using an existing bucket, the location of the bucket hasto be the same location yourVertex AI TensorBoard instance was created in.

GCS_BUCKET_NAME="PROJECT_ID-tensorboard-logs-LOCATION_ID"gcloudstoragebucketscreate"gs://${GCS_BUCKET_NAME}"--location=LOCATION_ID

ReplaceLOCATION_ID with the location that yourVertex AI TensorBoard instance was created in, for exampleus-central1.

GCS_BUCKET_NAME is used tocreate a custom training job with REST.

Create a Vertex AI TensorBoard instance

A Vertex AI TensorBoard instance, which is a regionalized resourcestoring your Vertex AI TensorBoard experiments, must be presentbefore experiments can be visualized. There are two options. You can eitheruse a default instance, or manually create one. Youcan create multipleinstances within a project and region, however most users only need a singleinstance.

Use the default Vertex AI TensorBoard instance

A default TensorBoard instance is automatically created when initializing aVertex AIexperiment. This backing TensorBoard is associated with the Vertex AI experiment and is used with all subsequent Vertex AI Experiments runs. Thetensorboard_resource_name can be reterived directly from the experiment. This is the easiest way to get started with Vertex AI TensorBoard and should meet most users needs.

fromgoogle.cloudimportaiplatformdefcreate_experiment_default_tensorboard_sample(experiment_name:str,experiment_description:str,project:str,location:str,):aiplatform.init(experiment=experiment_name,experiment_description=experiment_description,project=project,location=location,)tensorboard=aiplatform.Experiment(experiment_name).get_backing_tensorboard_resource()print(f"Tensorboard resource name:{tensorboard.name}")

Manually create a Vertex AI TensorBoard instance

You can manually create a Vertex AI TensorBoard. This is useful for users more comfortable with the Google Cloud console, users that need a CMEK enabled TensorBoard (seeCMEK), or users who want to use multiple TensorBoards. This instance can then be specified directly when initializing a Vertex AI experiment, starting an Experiment Run, or configuring thetraining code.

defcreate_tensorboard_sample(project:str,location:str,display_name:Optional[str]=None,):aiplatform.init(project=project,location=location)tensorboard=aiplatform.Tensorboard.create(display_name=display_name,project=project,location=location,)aiplatform.init(project=project,location=location,experiment_tensorboard=tensorboard)returntensorboard

resource"google_vertex_ai_tensorboard""default"{display_name="vertex-ai-tensorboard-sample-name"region="us-central1"}

Delete a TensorBoard instance

Deleting a TensorBoard instance deletes that TensorBoard and all associatedTensorBoard experiments and TensorBoard runs. TheVertex AI Experiments the instance is associated with isn't deleted.

To delete a Vertex AI Experiments and it's associatedVertex AI TensorBoard experiments, seeDelete an experiment.

def delete_tensorboard_instance_sample(    tensorboard_resource_name: str,    project: str,    location: str,):    aiplatform.init(project=project, location=location)    tensorboard = aiplatform.Tensorboard(        tensorboard_name=tensorboard_resource_name    )    tensorboard.delete()

Relevant terms

These terms, "TensorBoard resource name", and "TensorBoard instance ID" arereferenced in numerous samples.

TensorBoard resource name

The TensorBoard Resource name is used to fully identify theVertex AI TensorBoard instance. The format is as follows:

projects/PROJECT_ID_OR_NUMBER/locations/REGION/tensorboards/TENSORBOARD_INSTANCE_ID

The TensorBoard resource name is printed in the log messages when createdusing gcloud CLI or Vertex AI SDK, or can be created byproviding the appropriate values for the placeholders.

fromgoogle.cloudimportaiplatformdefget_experiment_backing_tensorboard_sample(experiment_name:str,project:str,location:str,):backing_tensorboard=aiplatform.Experiment(project=project,location=location,experiment_name=experiment_name).get_backing_tensorboard_resource()returnbacking_tensorboard.name

TensorBoard instance ID

The TensorBoard instance ID is a generated ID value associated with aTensorBoard instance. To find theTENSORBOARD_INSTANCE_ID, go to theExperiments page Vertex AI section of the Google Cloud console, andselect theTensorBoard Instances tab.

You can also retrieve the instance ID from theTensorBoard resource name.