def sample_create_hyperparameter_tuning_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
hyperparameter_tuning_job = aiplatform_v1beta1.HyperparameterTuningJob()
hyperparameter_tuning_job.display_name = "display_name_value"
hyperparameter_tuning_job.study_spec.metrics.metric_id = "metric_id_value"
hyperparameter_tuning_job.study_spec.metrics.goal = "MINIMIZE"
hyperparameter_tuning_job.study_spec.parameters.double_value_spec.min_value = 0.96
hyperparameter_tuning_job.study_spec.parameters.double_value_spec.max_value = 0.962
hyperparameter_tuning_job.study_spec.parameters.parameter_id = "parameter_id_value"
hyperparameter_tuning_job.max_trial_count = 1609
hyperparameter_tuning_job.parallel_trial_count = 2128
hyperparameter_tuning_job.trial_job_spec.worker_pool_specs.container_spec.image_uri = "image_uri_value"
request = aiplatform_v1beta1.CreateHyperparameterTuningJobRequest(
parent="parent_value",
hyperparameter_tuning_job=hyperparameter_tuning_job,
)
# Make the request
response = client.create_hyperparameter_tuning_job(request=request)
# Handle the response
print(response)
def sample_cancel_custom_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.CancelCustomJobRequest(name="name_value", )
# Make the request
client.cancel_custom_job(request=request)
def sample_cancel_hyperparameter_tuning_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.CancelHyperparameterTuningJobRequest(
name="name_value", )
# Make the request
client.cancel_hyperparameter_tuning_job(request=request)
def sample_cancel_data_labeling_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.CancelDataLabelingJobRequest(
name="name_value", )
# Make the request
client.cancel_data_labeling_job(request=request)
def sample_resume_model_deployment_monitoring_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.ResumeModelDeploymentMonitoringJobRequest(
name="name_value", )
# Make the request
client.resume_model_deployment_monitoring_job(request=request)
def sample_get_custom_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.GetCustomJobRequest(name="name_value", )
# Make the request
response = client.get_custom_job(request=request)
# Handle the response
print(response)
def sample_list_custom_jobs():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.ListCustomJobsRequest(parent="parent_value", )
# Make the request
page_result = client.list_custom_jobs(request=request)
# Handle the response
for response in page_result:
print(response)
def sample_get_hyperparameter_tuning_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.GetHyperparameterTuningJobRequest(
name="name_value", )
# Make the request
response = client.get_hyperparameter_tuning_job(request=request)
# Handle the response
print(response)
def sample_get_model_deployment_monitoring_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.GetModelDeploymentMonitoringJobRequest(
name="name_value", )
# Make the request
response = client.get_model_deployment_monitoring_job(request=request)
# Handle the response
print(response)
def sample_search_model_deployment_monitoring_stats_anomalies():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.SearchModelDeploymentMonitoringStatsAnomaliesRequest(
model_deployment_monitoring_job="model_deployment_monitoring_job_value",
deployed_model_id="deployed_model_id_value",
)
# Make the request
page_result = client.search_model_deployment_monitoring_stats_anomalies(request=request)
# Handle the response
for response in page_result:
print(response)
def sample_delete_custom_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
request = aiplatform_v1beta1.DeleteCustomJobRequest(name="name_value", )
# Make the request
operation = client.delete_custom_job(request=request)
print("Waiting for operation to complete...")
response = operation.result()
# Handle the response
print(response)
def sample_create_model_deployment_monitoring_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
model_deployment_monitoring_job = aiplatform_v1beta1.ModelDeploymentMonitoringJob()
model_deployment_monitoring_job.display_name = "display_name_value"
model_deployment_monitoring_job.endpoint = "endpoint_value"
request = aiplatform_v1beta1.CreateModelDeploymentMonitoringJobRequest(
parent="parent_value",
model_deployment_monitoring_job=model_deployment_monitoring_job,
)
# Make the request
response = client.create_model_deployment_monitoring_job(request=request)
# Handle the response
print(response)
def sample_create_custom_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
custom_job = aiplatform_v1beta1.CustomJob()
custom_job.display_name = "display_name_value"
custom_job.job_spec.worker_pool_specs.container_spec.image_uri = "image_uri_value"
request = aiplatform_v1beta1.CreateCustomJobRequest(
parent="parent_value",
custom_job=custom_job,
)
# Make the request
response = client.create_custom_job(request=request)
# Handle the response
print(response)
def create_batch_prediction_job_bigquery_sample(
project: str,
display_name: str,
model_name: str,
instances_format: str,
bigquery_source_input_uri: str,
predictions_format: str,
bigquery_destination_output_uri: str,
location: str = "us-central1",
api_endpoint: str = "us-central1-aiplatform.googleapis.com",
):
# The AI Platform services require regional API endpoints.
client_options = {"api_endpoint": api_endpoint}
# Initialize client that will be used to create and send requests.
# This client only needs to be created once, and can be reused for multiple requests.
client = aiplatform_v1beta1.JobServiceClient(client_options=client_options)
model_parameters_dict = {}
model_parameters = json_format.ParseDict(model_parameters_dict, Value())
batch_prediction_job = {
"display_name": display_name,
# Format: 'projects/{project}/locations/{location}/models/{model_id}'
"model": model_name,
"model_parameters": model_parameters,
"input_config": {
"instances_format": instances_format,
"bigquery_source": {
"input_uri": bigquery_source_input_uri
},
},
"output_config": {
"predictions_format": predictions_format,
"bigquery_destination": {
"output_uri": bigquery_destination_output_uri
},
},
# optional
"generate_explanation": True,
}
parent = f"projects/{project}/locations/{location}"
response = client.create_batch_prediction_job(
parent=parent, batch_prediction_job=batch_prediction_job)
print("response:", response)
def sample_update_model_deployment_monitoring_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
model_deployment_monitoring_job = aiplatform_v1beta1.ModelDeploymentMonitoringJob(
)
model_deployment_monitoring_job.display_name = "display_name_value"
model_deployment_monitoring_job.endpoint = "endpoint_value"
request = aiplatform_v1beta1.UpdateModelDeploymentMonitoringJobRequest(
model_deployment_monitoring_job=model_deployment_monitoring_job, )
# Make the request
operation = client.update_model_deployment_monitoring_job(request=request)
print("Waiting for operation to complete...")
response = operation.result()
# Handle the response
print(response)
def sample_create_data_labeling_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
data_labeling_job = aiplatform_v1beta1.DataLabelingJob()
data_labeling_job.display_name = "display_name_value"
data_labeling_job.datasets = ['datasets_value_1', 'datasets_value_2']
data_labeling_job.labeler_count = 1375
data_labeling_job.instruction_uri = "instruction_uri_value"
data_labeling_job.inputs_schema_uri = "inputs_schema_uri_value"
data_labeling_job.inputs.null_value = "NULL_VALUE"
request = aiplatform_v1beta1.CreateDataLabelingJobRequest(
parent="parent_value",
data_labeling_job=data_labeling_job,
)
# Make the request
response = client.create_data_labeling_job(request=request)
# Handle the response
print(response)
def create_batch_prediction_job_tabular_forecasting_sample(
project: str,
display_name: str,
model_name: str,
gcs_source_uri: str,
gcs_destination_output_uri_prefix: str,
predictions_format: str,
location: str = "us-central1",
api_endpoint: str = "us-central1-aiplatform.googleapis.com",
):
# The AI Platform services require regional API endpoints.
client_options = {"api_endpoint": api_endpoint}
# Initialize client that will be used to create and send requests.
# This client only needs to be created once, and can be reused for multiple requests.
client = aiplatform_v1beta1.JobServiceClient(client_options=client_options)
batch_prediction_job = {
"display_name": display_name,
# Format: 'projects/{project}/locations/{location}/models/{model_id}'
"model": model_name,
"input_config": {
"instances_format": predictions_format,
"gcs_source": {
"uris": [gcs_source_uri]
},
},
"output_config": {
"predictions_format": predictions_format,
"gcs_destination": {
"output_uri_prefix": gcs_destination_output_uri_prefix
},
},
}
parent = f"projects/{project}/locations/{location}"
response = client.create_batch_prediction_job(
parent=parent, batch_prediction_job=batch_prediction_job)
print("response:", response)
def sample_create_batch_prediction_job():
# Create a client
client = aiplatform_v1beta1.JobServiceClient()
# Initialize request argument(s)
batch_prediction_job = aiplatform_v1beta1.BatchPredictionJob()
batch_prediction_job.display_name = "display_name_value"
batch_prediction_job.input_config.gcs_source.uris = [
'uris_value_1', 'uris_value_2'
]
batch_prediction_job.input_config.instances_format = "instances_format_value"
batch_prediction_job.output_config.gcs_destination.output_uri_prefix = "output_uri_prefix_value"
batch_prediction_job.output_config.predictions_format = "predictions_format_value"
request = aiplatform_v1beta1.CreateBatchPredictionJobRequest(
parent="parent_value",
batch_prediction_job=batch_prediction_job,
)
# Make the request
response = client.create_batch_prediction_job(request=request)
# Handle the response
print(response)
def execute_notebook(args):
client_info = gapic_v1.client_info.ClientInfo(
user_agent="google-cloud-pipeline-components", )
client_notebooks = notebooks.NotebookServiceClient(client_info=client_info)
client_vertexai_jobs = vertex_ai_beta.JobServiceClient(
client_options={
'api_endpoint': f'{args.location}-aiplatform.googleapis.com'
},
client_info=client_info)
execution_parent = f'projects/{args.project}/locations/{args.location}'
execution_fullname = f'{execution_parent}/executions/{args.execution_id}'
execution_template = build_execution_template(args)
gcp_resources = ''
try:
print('Try create_execution()...')
execution_create_operation = client_notebooks.create_execution(
parent=execution_parent,
execution_id=args.execution_id,
execution=execution_template)
gcp_resources = json.dumps({
"resources": [{
"resourceType":
'type.googleapis.com/google.cloud.notebooks.v1.Execution',
"resourceUri": execution_fullname
}]
})
except Exception as e:
response = build_response(error=f'create_execution() failed: {e}')
handle_error(args.fail_pipeline, response)
return response
# Gets initial execution
try:
print('Try get_execution()...')
execution = client_notebooks.get_execution(name=execution_fullname)
except Exception as e:
response = build_response(error=f'get_execution() failed: {e}')
handle_error(args.fail_pipeline, response)
return response
if not args.block_pipeline:
print('Not blocking pipeline...')
return build_response(
state=Execution.State(execution.state).name,
output_notebook_file=execution.output_notebook_file,
gcp_resources=gcp_resources)
# Waits for execution to finish.
print('Blocking pipeline...')
execution_state = ''
execution_job_uri = ''
while True:
try:
execution = client_notebooks.get_execution(name=execution_fullname)
execution_state = getattr(execution, 'state', '')
print(
f'execution.state is {Execution.State(execution_state).name}')
except Exception as e:
response = build_response(
error=f'get_execution() for blocking pipeline failed: {e}')
handle_error(args.fail_pipeline, response)
return response
# Job URI is not available when state is INITIALIZING.
if execution_state in _STATES_JOB_URI and not execution_job_uri:
execution_job_uri = getattr(execution, 'job_uri', '')
print(f'execution.job_uri is {execution_job_uri}')
if execution_state in _STATES_COMPLETED:
break
time.sleep(30)
# For some reason, execution.get and job.get might not return the same state
# so we check if we can get the error message using the AI Plaform API.
# It is only possible if there is job_uri available though.
if execution_state in _STATES_ERROR:
if execution_job_uri:
while True:
try:
custom_job = client_vertexai_jobs.get_custom_job(
name=execution_job_uri)
except Exception as e:
response = build_response(
error=f'get_custom_job() failed: {e}')
handle_error(args.fail_pipeline, response)
return response
custom_job_state = getattr(custom_job, 'state', None)
if custom_job_state in _STATES_COMPLETED:
break
time.sleep(30)
# == to `if state in _JOB_ERROR_STATES`
custom_job_error = getattr(custom_job, 'error', None)
if custom_job_error:
response = build_response(
error=
f'Error {custom_job_error.code}: {custom_job_error.message}'
)
handle_error(args.fail_pipeline, (None, response))
return response
# The job might be successful but we need to address that the execution
# had a problem. The previous loop was in hope to find the error message,
# we didn't have any so we return the execution state as the message.
response = build_response(
error=
f'Execution finished with state: {Execution.State(execution_state).name}'
)
handle_error(args.fail_pipeline, (None, response))
return response
return build_response(state=Execution.State(execution_state).name,
output_notebook_file=execution.output_notebook_file,
gcp_resources=gcp_resources)
