def test_deploy_model(
sklearn_training_job,
sagemaker_session,
cpu_instance_type,
sklearn_latest_version,
sklearn_latest_py_version,
):
endpoint_name = "test-sklearn-deploy-model-{}".format(
sagemaker_timestamp())
with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session):
desc = sagemaker_session.sagemaker_client.describe_training_job(
TrainingJobName=sklearn_training_job)
model_data = desc["ModelArtifacts"]["S3ModelArtifacts"]
script_path = os.path.join(DATA_DIR, "sklearn_mnist", "mnist.py")
model = SKLearnModel(
model_data,
"SageMakerRole",
entry_point=script_path,
framework_version=sklearn_latest_version,
sagemaker_session=sagemaker_session,
)
predictor = model.deploy(1,
cpu_instance_type,
endpoint_name=endpoint_name)
_predict_and_assert(predictor)
def test_model(sagemaker_session):
model = SKLearnModel("s3://some/data.tar.gz",
role=ROLE,
entry_point=SCRIPT_PATH,
sagemaker_session=sagemaker_session)
predictor = model.deploy(1, CPU)
assert isinstance(predictor, SKLearnPredictor)
def main():
sagemaker_session = LocalSession()
sagemaker_session.config = {'local': {'local_code': True}}
dummy_model_file = Path("dummy.model")
dummy_model_file.touch()
with tarfile.open("model.tar.gz", "w:gz") as tar:
tar.add(dummy_model_file.as_posix())
# For local training a dummy role will be sufficient
role = DUMMY_IAM_ROLE
model = SKLearnModel(role=role,
model_data='file://./model.tar.gz',
framework_version='0.23-1',
py_version='py3',
source_dir='code',
entry_point='inference.py')
print('Deploying endpoint in local mode')
print(
'Note: if launching for the first time in local mode, container image download might take a few minutes to complete.'
)
predictor = model.deploy(
initial_instance_count=1,
instance_type='local',
)
do_inference_on_local_endpoint(predictor)
print('About to delete the endpoint to stop paying (if in cloud mode).')
predictor.delete_endpoint(predictor.endpoint_name)
def test_deploy_model(sklearn_training_job, sagemaker_session):
endpoint_name = 'test-sklearn-deploy-model-{}'.format(sagemaker_timestamp())
with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session):
desc = sagemaker_session.sagemaker_client.describe_training_job(TrainingJobName=sklearn_training_job)
model_data = desc['ModelArtifacts']['S3ModelArtifacts']
script_path = os.path.join(DATA_DIR, 'sklearn_mnist', 'mnist.py')
model = SKLearnModel(model_data, 'SageMakerRole', entry_point=script_path, sagemaker_session=sagemaker_session)
predictor = model.deploy(1, "ml.m4.xlarge", endpoint_name=endpoint_name)
_predict_and_assert(predictor)
def test_model_custom_serialization(sagemaker_session, sklearn_version):
model = SKLearnModel(
"s3://some/data.tar.gz",
role=ROLE,
entry_point=SCRIPT_PATH,
framework_version=sklearn_version,
sagemaker_session=sagemaker_session,
)
custom_serializer = Mock()
custom_deserializer = Mock()
predictor = model.deploy(
1,
CPU,
serializer=custom_serializer,
deserializer=custom_deserializer,
)
assert isinstance(predictor, SKLearnPredictor)
assert predictor.serializer is custom_serializer
assert predictor.deserializer is custom_deserializer
def main():
# Prepare data for model inference - we use the Boston housing dataset
print('Preparing data for model inference')
data = fetch_california_housing()
X_train, X_test, y_train, y_test = train_test_split(data.data,
data.target,
test_size=0.25,
random_state=42)
# we don't train a model, so we will need only the testing data
testX = pd.DataFrame(X_test, columns=data.feature_names)
# Download a pre-trained model file
print('Downloading a pre-trained model file')
s3.download_file(
'aws-ml-blog',
'artifacts/scikit_learn_bring_your_own_model/model.joblib',
'model.joblib')
# Creating a model.tar.gz file
tar = tarfile.open('model.tar.gz', 'w:gz')
tar.add('model.joblib')
tar.close()
model = SKLearnModel(role=DUMMY_IAM_ROLE,
model_data='file://./model.tar.gz',
framework_version='0.23-1',
py_version='py3',
source_dir='code',
entry_point='inference.py')
print('Deploying endpoint in local mode')
predictor = model.deploy(initial_instance_count=1, instance_type='local')
predictions = predictor.predict(testX[data.feature_names].head(5))
print(f"Predictions: {predictions}")
print('About to delete the endpoint to stop paying (if in cloud mode).')
predictor.delete_endpoint(predictor.endpoint_name)
entry_point='entry_point.py',
source_dir=str(Path(current_folder, '../containers/model/src').resolve()),
dependencies=[str(Path(current_folder, '../package/package').resolve())],
image=config.ECR_IMAGE,
code_location='s3://' +
str(Path(config.S3_BUCKET, config.OUTPUTS_S3_PREFIX)))
# %% [markdown]
# Calling `deploy` will start a container to host the model.
# You can expect this step to take approximately 5 minutes.
# %%
model.deploy(endpoint_name=explainer_name,
instance_type='ml.c5.xlarge',
initial_instance_count=1,
tags=[{
'Key': config.TAG_KEY,
'Value': config.SOLUTION_PREFIX
}])
# %% [markdown]
# When you're trying to update the model for development purposes, but
# experiencing issues because the model/endpoint-config/endpoint already
# exists, you can delete the existing model/endpoint-config/endpoint by
# uncommenting and running the following commands:
# %%
# sagemaker_client.delete_endpoint(EndpointName=explainer_name)
# sagemaker_client.delete_endpoint_config(EndpointConfigName=explainer_name)
# sagemaker_client.delete_model(ModelName=explainer_name)