def test_stop_tuning_job(sagemaker_session, tuner):
sagemaker_session.stop_tuning_job = Mock(name='stop_hyper_parameter_tuning_job')
tuner.latest_tuning_job = _TuningJob(sagemaker_session, JOB_NAME)
tuner.stop_tuning_job()
sagemaker_session.stop_tuning_job.assert_called_once_with(name=JOB_NAME)
def test_wait(tuner):
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
tuner.estimator.sagemaker_session.wait_for_tuning_job = Mock(name='wait_for_tuning_job')
tuner.wait()
tuner.estimator.sagemaker_session.wait_for_tuning_job.assert_called_once_with(JOB_NAME)
def test_stop_tuning_job(sagemaker_session, tuner):
sagemaker_session.stop_tuning_job = Mock(name='stop_hyper_parameter_tuning_job')
tuner.latest_tuning_job = _TuningJob(sagemaker_session, JOB_NAME)
tuner.stop_tuning_job()
sagemaker_session.stop_tuning_job.assert_called_once_with(name=JOB_NAME)
def test_deploy_default(tuner):
returned_training_job_description = {
'AlgorithmSpecification': {
'TrainingInputMode': 'File',
'TrainingImage': IMAGE_NAME
},
'HyperParameters': {
'sagemaker_submit_directory': '"s3://some/sourcedir.tar.gz"',
'checkpoint_path': '"s3://other/1508872349"',
'sagemaker_program': '"iris-dnn-classifier.py"',
'sagemaker_enable_cloudwatch_metrics': 'false',
'sagemaker_container_log_level': '"logging.INFO"',
'sagemaker_job_name': '"neo"',
'training_steps': '100',
'_tuning_objective_metric': 'Validation-accuracy',
},
'RoleArn': ROLE,
'ResourceConfig': {
'VolumeSizeInGB': 30,
'InstanceCount': 1,
'InstanceType': 'ml.c4.xlarge'
},
'StoppingCondition': {
'MaxRuntimeInSeconds': 24 * 60 * 60
},
'TrainingJobName': 'neo',
'TrainingJobStatus': 'Completed',
'OutputDataConfig': {
'KmsKeyId': '',
'S3OutputPath': 's3://place/output/neo'
},
'TrainingJobOutput': {
'S3TrainingJobOutput': 's3://here/output.tar.gz'
},
'ModelArtifacts': {
'S3ModelArtifacts': MODEL_DATA
}
}
tuning_job_description = {'BestTrainingJob': {'TrainingJobName': JOB_NAME}}
tuner.estimator.sagemaker_session.sagemaker_client.describe_training_job = \
Mock(name='describe_training_job', return_value=returned_training_job_description)
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job = Mock(
name='describe_hyper_parameter_tuning_job', return_value=tuning_job_description)
tuner.estimator.sagemaker_session.log_for_jobs = Mock(name='log_for_jobs')
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
predictor = tuner.deploy(TRAIN_INSTANCE_COUNT, TRAIN_INSTANCE_TYPE)
tuner.estimator.sagemaker_session.create_model.assert_called_once()
args = tuner.estimator.sagemaker_session.create_model.call_args[0]
assert args[0].startswith(IMAGE_NAME)
assert args[1] == ROLE
assert args[2]['Image'] == IMAGE_NAME
assert args[2]['ModelDataUrl'] == MODEL_DATA
assert isinstance(predictor, RealTimePredictor)
assert predictor.endpoint.startswith(JOB_NAME)
assert predictor.sagemaker_session == tuner.estimator.sagemaker_session
def test_deploy_default(tuner):
returned_training_job_description = {
'AlgorithmSpecification': {
'TrainingInputMode': 'File',
'TrainingImage': IMAGE_NAME
},
'HyperParameters': {
'sagemaker_submit_directory': '"s3://some/sourcedir.tar.gz"',
'checkpoint_path': '"s3://other/1508872349"',
'sagemaker_program': '"iris-dnn-classifier.py"',
'sagemaker_enable_cloudwatch_metrics': 'false',
'sagemaker_container_log_level': '"logging.INFO"',
'sagemaker_job_name': '"neo"',
'training_steps': '100',
'_tuning_objective_metric': 'Validation-accuracy',
},
'RoleArn': ROLE,
'ResourceConfig': {
'VolumeSizeInGB': 30,
'InstanceCount': 1,
'InstanceType': 'ml.c4.xlarge'
},
'StoppingCondition': {
'MaxRuntimeInSeconds': 24 * 60 * 60
},
'TrainingJobName': 'neo',
'TrainingJobStatus': 'Completed',
'OutputDataConfig': {
'KmsKeyId': '',
'S3OutputPath': 's3://place/output/neo'
},
'TrainingJobOutput': {
'S3TrainingJobOutput': 's3://here/output.tar.gz'
},
'ModelArtifacts': {
'S3ModelArtifacts': MODEL_DATA
}
}
tuning_job_description = {'BestTrainingJob': {'TrainingJobName': JOB_NAME}}
tuner.estimator.sagemaker_session.sagemaker_client.describe_training_job = \
Mock(name='describe_training_job', return_value=returned_training_job_description)
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job = Mock(
name='describe_hyper_parameter_tuning_job', return_value=tuning_job_description)
tuner.estimator.sagemaker_session.log_for_jobs = Mock(name='log_for_jobs')
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
predictor = tuner.deploy(TRAIN_INSTANCE_COUNT, TRAIN_INSTANCE_TYPE)
tuner.estimator.sagemaker_session.create_model.assert_called_once()
args = tuner.estimator.sagemaker_session.create_model.call_args[0]
assert args[0].startswith(IMAGE_NAME)
assert args[1] == ROLE
assert args[2]['Image'] == IMAGE_NAME
assert args[2]['ModelDataUrl'] == MODEL_DATA
assert isinstance(predictor, RealTimePredictor)
assert predictor.endpoint.startswith(JOB_NAME)
assert predictor.sagemaker_session == tuner.estimator.sagemaker_session
def test_wait(tuner):
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
tuner.estimator.sagemaker_session.wait_for_tuning_job = Mock(name='wait_for_tuning_job')
tuner.wait()
tuner.estimator.sagemaker_session.wait_for_tuning_job.assert_called_once_with(JOB_NAME)
def test_tuning_job_wait(sagemaker_session):
sagemaker_session.wait_for_tuning_job = Mock(name='wait_for_tuning_job')
tuning_job = _TuningJob(sagemaker_session, JOB_NAME)
tuning_job.wait()
sagemaker_session.wait_for_tuning_job.assert_called_once_with(JOB_NAME)
def test_tuning_job_wait(sagemaker_session):
sagemaker_session.wait_for_tuning_job = Mock(name='wait_for_tuning_job')
tuning_job = _TuningJob(sagemaker_session, JOB_NAME)
tuning_job.wait()
sagemaker_session.wait_for_tuning_job.assert_called_once_with(JOB_NAME)
def test_deploy_default(tuner):
returned_training_job_description = {
"AlgorithmSpecification": {
"TrainingInputMode": "File",
"TrainingImage": IMAGE_NAME,
"MetricDefinitions": METRIC_DEFINITIONS,
},
"HyperParameters": {
"sagemaker_submit_directory": '"s3://some/sourcedir.tar.gz"',
"checkpoint_path": '"s3://other/1508872349"',
"sagemaker_program": '"iris-dnn-classifier.py"',
"sagemaker_enable_cloudwatch_metrics": "false",
"sagemaker_container_log_level": '"logging.INFO"',
"sagemaker_job_name": '"neo"',
"training_steps": "100",
"_tuning_objective_metric": "Validation-accuracy",
},
"RoleArn": ROLE,
"ResourceConfig": {
"VolumeSizeInGB": 30,
"InstanceCount": 1,
"InstanceType": "ml.c4.xlarge",
},
"StoppingCondition": {"MaxRuntimeInSeconds": 24 * 60 * 60},
"TrainingJobName": "neo",
"TrainingJobStatus": "Completed",
"TrainingJobArn": "arn:aws:sagemaker:us-west-2:336:training-job/neo",
"OutputDataConfig": {"KmsKeyId": "", "S3OutputPath": "s3://place/output/neo"},
"TrainingJobOutput": {"S3TrainingJobOutput": "s3://here/output.tar.gz"},
"ModelArtifacts": {"S3ModelArtifacts": MODEL_DATA},
}
tuning_job_description = {"BestTrainingJob": {"TrainingJobName": JOB_NAME}}
returned_list_tags = {"Tags": [{"Key": "TagtestKey", "Value": "TagtestValue"}]}
tuner.estimator.sagemaker_session.sagemaker_client.describe_training_job = Mock(
name="describe_training_job", return_value=returned_training_job_description
)
tuner.estimator.sagemaker_session.sagemaker_client.list_tags = Mock(
name="list_tags", return_value=returned_list_tags
)
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job = Mock(
name="describe_hyper_parameter_tuning_job", return_value=tuning_job_description
)
tuner.estimator.sagemaker_session.log_for_jobs = Mock(name="log_for_jobs")
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
predictor = tuner.deploy(TRAIN_INSTANCE_COUNT, TRAIN_INSTANCE_TYPE)
tuner.estimator.sagemaker_session.create_model.assert_called_once()
args = tuner.estimator.sagemaker_session.create_model.call_args[0]
assert args[0] == "neo"
assert args[1] == ROLE
assert args[2]["Image"] == IMAGE_NAME
assert args[2]["ModelDataUrl"] == MODEL_DATA
assert isinstance(predictor, RealTimePredictor)
assert predictor.endpoint.startswith(JOB_NAME)
assert predictor.sagemaker_session == tuner.estimator.sagemaker_session
def test_start_new(tuner, sagemaker_session):
tuning_job = _TuningJob(sagemaker_session, JOB_NAME)
tuner.static_hyperparameters = {}
started_tuning_job = tuning_job.start_new(tuner, INPUTS)
assert started_tuning_job.sagemaker_session == sagemaker_session
sagemaker_session.tune.assert_called_once()
def test_start_new(tuner, sagemaker_session):
tuning_job = _TuningJob(sagemaker_session, JOB_NAME)
tuner.static_hyperparameters = {}
started_tuning_job = tuning_job.start_new(tuner, INPUTS)
assert started_tuning_job.sagemaker_session == sagemaker_session
sagemaker_session.tune.assert_called_once()
def test_delete_endpoint(tuner):
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
tuning_job_description = {'BestTrainingJob': {'TrainingJobName': JOB_NAME}}
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job = Mock(
name='describe_hyper_parameter_tuning_job', return_value=tuning_job_description)
tuner.delete_endpoint()
tuner.sagemaker_session.delete_endpoint.assert_called_with(JOB_NAME)
def test_delete_endpoint(tuner):
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
tuning_job_description = {'BestTrainingJob': {'TrainingJobName': JOB_NAME}}
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job = Mock(
name='describe_hyper_parameter_tuning_job', return_value=tuning_job_description)
tuner.delete_endpoint()
tuner.sagemaker_session.delete_endpoint.assert_called_with(JOB_NAME)
def test_best_tuning_job(tuner):
tuning_job_description = {'BestTrainingJob': {'TrainingJobName': JOB_NAME}}
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job = Mock(
name='describe_hyper_parameter_tuning_job', return_value=tuning_job_description)
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
best_training_job = tuner.best_training_job()
assert best_training_job == JOB_NAME
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job.assert_called_once_with(
HyperParameterTuningJobName=JOB_NAME)
def test_best_tuning_job(tuner):
tuning_job_description = {'BestTrainingJob': {'TrainingJobName': JOB_NAME}}
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job = Mock(
name='describe_hyper_parameter_tuning_job', return_value=tuning_job_description)
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
best_training_job = tuner.best_training_job()
assert best_training_job == JOB_NAME
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job.assert_called_once_with(
HyperParameterTuningJobName=JOB_NAME)
def test_best_tuning_job_no_best_job(tuner):
tuning_job_description = {'BestTrainingJob': {'Mock': None}}
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job = Mock(
name='describe_hyper_parameter_tuning_job', return_value=tuning_job_description)
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
with pytest.raises(Exception) as e:
tuner.best_training_job()
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job.assert_called_once_with(
HyperParameterTuningJobName=JOB_NAME)
assert 'Best training job not available for tuning job:' in str(e)
def test_best_tuning_job_no_best_job(tuner):
tuning_job_description = {'BestTrainingJob': {'Mock': None}}
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job = Mock(
name='describe_hyper_parameter_tuning_job', return_value=tuning_job_description)
tuner.latest_tuning_job = _TuningJob(tuner.estimator.sagemaker_session, JOB_NAME)
with pytest.raises(Exception) as e:
tuner.best_training_job()
tuner.estimator.sagemaker_session.sagemaker_client.describe_hyper_parameter_tuning_job.assert_called_once_with(
HyperParameterTuningJobName=JOB_NAME)
assert 'Best training job not available for tuning job:' in str(e)
def test_analytics(tuner):
tuner.latest_tuning_job = _TuningJob(tuner.sagemaker_session, 'testjob')
tuner_analytics = tuner.analytics()
assert tuner_analytics is not None
assert tuner_analytics.name.find('testjob') > -1
def test_stop(sagemaker_session):
tuning_job = _TuningJob(sagemaker_session, JOB_NAME)
tuning_job.stop()
sagemaker_session.stop_tuning_job.assert_called_once_with(name=JOB_NAME)
def test_analytics(tuner):
tuner.latest_tuning_job = _TuningJob(tuner.sagemaker_session, 'testjob')
tuner_analytics = tuner.analytics()
assert tuner_analytics is not None
assert tuner_analytics.name.find('testjob') > -1
def test_stop(sagemaker_session):
tuning_job = _TuningJob(sagemaker_session, JOB_NAME)
tuning_job.stop()
sagemaker_session.stop_tuning_job.assert_called_once_with(name=JOB_NAME)
