def __init__(self,
state_id,
model,
model_name=None,
instance_type=None,
tags=None,
**kwargs):
"""
Args:
state_id (str): State name whose length **must be** less than or equal to 128 unicode characters. State names **must be** unique within the scope of the whole state machine.
model (sagemaker.model.Model): The SageMaker model to use in the ModelStep. If :py:class:`TrainingStep` was used to train the model and saving the model is the next step in the workflow, the output of :py:func:`TrainingStep.get_expected_model()` can be passed here.
model_name (str or Placeholder, optional): Specify a model name, this is required for creating the model. We recommend to use :py:class:`~stepfunctions.inputs.ExecutionInput` placeholder collection to pass the value dynamically in each execution.
instance_type (str, optional): The EC2 instance type to deploy this Model to. For example, 'ml.p2.xlarge'.
tags (list[dict] or Placeholders, optional): `List of tags <https://docs.aws.amazon.com/sagemaker/latest/dg/API_Tag.html>`_ to associate with the resource.
parameters(dict, optional): The value of this field is merged with other arguments to become the request payload for SageMaker `CreateModel <https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_CreateModel.html>`_. (Default: None)
You can use `parameters` to override the value provided by other arguments and specify any field's value dynamically using `Placeholders <https://aws-step-functions-data-science-sdk.readthedocs.io/en/stable/placeholders.html?highlight=placeholder#stepfunctions.inputs.Placeholder>`_.
"""
if isinstance(model, FrameworkModel):
model_parameters = model_config(model=model,
instance_type=instance_type,
role=model.role,
image_uri=model.image_uri)
if model_name:
model_parameters['ModelName'] = model_name
elif isinstance(model, Model):
model_parameters = {
'ExecutionRoleArn': model.role,
'ModelName': model_name or model.name,
'PrimaryContainer': {
'Environment': model.env,
'Image': model.image_uri,
'ModelDataUrl': model.model_data
}
}
else:
raise ValueError(
"Expected 'model' parameter to be of type 'sagemaker.model.Model', but received type '{}'"
.format(type(model).__name__))
if 'S3Operations' in model_parameters:
del model_parameters['S3Operations']
if tags:
model_parameters['Tags'] = tags if isinstance(
tags, Placeholder) else tags_dict_to_kv_list(tags)
if Field.Parameters.value in kwargs and isinstance(
kwargs[Field.Parameters.value], dict):
# Update model parameters with input parameters
merge_dicts(model_parameters, kwargs[Field.Parameters.value])
kwargs[Field.Parameters.value] = model_parameters
"""
Example resource arn: arn:aws:states:::sagemaker:createModel
"""
kwargs[Field.Resource.value] = get_service_integration_arn(
SAGEMAKER_SERVICE_NAME, SageMakerApi.CreateModel)
super(ModelStep, self).__init__(state_id, **kwargs)
def test_byo_model_config(sagemaker_session):
byo_model = model.Model(
model_data="{{ model_data }}",
image="{{ image }}",
role="{{ role }}",
env={"{{ key }}": "{{ value }}"},
name="model",
sagemaker_session=sagemaker_session,
)
config = airflow.model_config(instance_type="ml.c4.xlarge",
model=byo_model)
expected_config = {
"ModelName": "model",
"PrimaryContainer": {
"Image": "{{ image }}",
"Environment": {
"{{ key }}": "{{ value }}"
},
"ModelDataUrl": "{{ model_data }}",
},
"ExecutionRoleArn": "{{ role }}",
}
assert config == expected_config
def __init__(self, state_id, model, model_name=None, instance_type=None, **kwargs):
"""
Args:
state_id (str): State name whose length **must be** less than or equal to 128 unicode characters. State names **must be** unique within the scope of the whole state machine.
model (sagemaker.model.Model): The SageMaker model to use in the ModelStep. If :py:class:`TrainingStep` was used to train the model and saving the model is the next step in the workflow, the output of :py:func:`TrainingStep.get_expected_model()` can be passed here.
model_name (str or Placeholder, optional): Specify a model name, this is required for creating the model. We recommend to use :py:class:`~stepfunctions.inputs.ExecutionInput` placeholder collection to pass the value dynamically in each execution.
instance_type (str, optional): The EC2 instance type to deploy this Model to. For example, 'ml.p2.xlarge'. This parameter is typically required when the estimator used is not an `Amazon built-in algorithm <https://docs.aws.amazon.com/sagemaker/latest/dg/algos.html>`_.
"""
if isinstance(model, FrameworkModel):
parameters = model_config(model=model, instance_type=instance_type, role=model.role, image=model.image)
if model_name:
parameters['ModelName'] = model_name
elif isinstance(model, Model):
parameters = {
'ExecutionRoleArn': model.role,
'ModelName': model_name or model.name,
'PrimaryContainer': {
'Environment': {},
'Image': model.image,
'ModelDataUrl': model.model_data
}
}
else:
raise ValueError("Expected 'model' parameter to be of type 'sagemaker.model.Model', but received type '{}'".format(type(model).__name__))
if 'S3Operations' in parameters:
del parameters['S3Operations']
kwargs[Field.Parameters.value] = parameters
kwargs[Field.Resource.value] = 'arn:aws:states:::sagemaker:createModel'
super(ModelStep, self).__init__(state_id, **kwargs)
def test_framework_model_config(sagemaker_session):
chainer_model = chainer.ChainerModel(
model_data="{{ model_data }}",
role="{{ role }}",
entry_point="{{ entry_point }}",
source_dir="{{ source_dir }}",
image=None,
py_version="py3",
framework_version="5.0.0",
model_server_workers="{{ model_server_worker }}",
sagemaker_session=sagemaker_session,
)
config = airflow.model_config(instance_type="ml.c4.xlarge",
model=chainer_model)
expected_config = {
"ModelName": "sagemaker-chainer-%s" % TIME_STAMP,
"PrimaryContainer": {
"Image":
"520713654638.dkr.ecr.us-west-2.amazonaws.com/sagemaker-chainer:5.0.0-cpu-py3",
"Environment": {
"SAGEMAKER_PROGRAM":
"{{ entry_point }}",
"SAGEMAKER_SUBMIT_DIRECTORY":
"s3://output/sagemaker-chainer-%s/source/sourcedir.tar.gz" %
TIME_STAMP,
"SAGEMAKER_ENABLE_CLOUDWATCH_METRICS":
"false",
"SAGEMAKER_CONTAINER_LOG_LEVEL":
"20",
"SAGEMAKER_REGION":
"us-west-2",
"SAGEMAKER_MODEL_SERVER_WORKERS":
"{{ model_server_worker }}",
},
"ModelDataUrl": "{{ model_data }}",
},
"ExecutionRoleArn": "{{ role }}",
"S3Operations": {
"S3Upload": [{
"Path":
"{{ source_dir }}",
"Bucket":
"output",
"Key":
"sagemaker-chainer-%s/source/sourcedir.tar.gz" % TIME_STAMP,
"Tar":
True,
}]
},
}
assert config == expected_config
def __init__(self,
state_id,
model,
model_name=None,
instance_type=None,
tags=None,
**kwargs):
"""
Args:
state_id (str): State name whose length **must be** less than or equal to 128 unicode characters. State names **must be** unique within the scope of the whole state machine.
model (sagemaker.model.Model): The SageMaker model to use in the ModelStep. If :py:class:`TrainingStep` was used to train the model and saving the model is the next step in the workflow, the output of :py:func:`TrainingStep.get_expected_model()` can be passed here.
model_name (str or Placeholder, optional): Specify a model name, this is required for creating the model. We recommend to use :py:class:`~stepfunctions.inputs.ExecutionInput` placeholder collection to pass the value dynamically in each execution.
instance_type (str, optional): The EC2 instance type to deploy this Model to. For example, 'ml.p2.xlarge'.
tags (list[dict], optional): `List to tags <https://docs.aws.amazon.com/sagemaker/latest/dg/API_Tag.html>`_ to associate with the resource.
"""
if isinstance(model, FrameworkModel):
parameters = model_config(model=model,
instance_type=instance_type,
role=model.role,
image_uri=model.image_uri)
if model_name:
parameters['ModelName'] = model_name
elif isinstance(model, Model):
parameters = {
'ExecutionRoleArn': model.role,
'ModelName': model_name or model.name,
'PrimaryContainer': {
'Environment': model.env,
'Image': model.image_uri,
'ModelDataUrl': model.model_data
}
}
else:
raise ValueError(
"Expected 'model' parameter to be of type 'sagemaker.model.Model', but received type '{}'"
.format(type(model).__name__))
if 'S3Operations' in parameters:
del parameters['S3Operations']
if tags:
parameters['Tags'] = tags_dict_to_kv_list(tags)
kwargs[Field.Parameters.value] = parameters
"""
Example resource arn: arn:aws:states:::sagemaker:createModel
"""
kwargs[Field.Resource.value] = get_service_integration_arn(
SAGEMAKER_SERVICE_NAME, SageMakerApi.CreateModel)
super(ModelStep, self).__init__(state_id, **kwargs)
def test_framework_model_config(sagemaker_session):
chainer_model = chainer.ChainerModel(
model_data="{{ model_data }}",
role="{{ role }}",
entry_point="{{ entry_point }}",
source_dir="{{ source_dir }}",
image=None,
py_version='py3',
framework_version='5.0.0',
model_server_workers="{{ model_server_worker }}",
sagemaker_session=sagemaker_session)
config = airflow.model_config(instance_type='ml.c4.xlarge',
model=chainer_model)
expected_config = {
'ModelName': "sagemaker-chainer-%s" % TIME_STAMP,
'PrimaryContainer': {
'Image':
'520713654638.dkr.ecr.us-west-2.amazonaws.com/sagemaker-chainer:5.0.0-cpu-py3',
'Environment': {
'SAGEMAKER_PROGRAM':
'{{ entry_point }}',
'SAGEMAKER_SUBMIT_DIRECTORY':
"s3://output/sagemaker-chainer-%s/source/sourcedir.tar.gz" %
TIME_STAMP,
'SAGEMAKER_ENABLE_CLOUDWATCH_METRICS':
'false',
'SAGEMAKER_CONTAINER_LOG_LEVEL':
'20',
'SAGEMAKER_REGION':
'us-west-2',
'SAGEMAKER_MODEL_SERVER_WORKERS':
'{{ model_server_worker }}'
},
'ModelDataUrl': '{{ model_data }}'
},
'ExecutionRoleArn': '{{ role }}',
'S3Operations': {
'S3Upload': [{
'Path':
'{{ source_dir }}',
'Bucket':
'output',
'Key':
"sagemaker-chainer-%s/source/sourcedir.tar.gz" % TIME_STAMP,
'Tar':
True
}]
}
}
assert config == expected_config
def _build_airflow_workflow(estimator,
instance_type,
inputs=None,
mini_batch_size=None):
training_config = sm_airflow.training_config(
estimator=estimator, inputs=inputs, mini_batch_size=mini_batch_size)
model = estimator.create_model()
assert model is not None
model_config = sm_airflow.model_config(instance_type, model)
assert model_config is not None
transform_config = sm_airflow.transform_config_from_estimator(
estimator=estimator,
task_id="transform_config",
task_type="training",
instance_count=SINGLE_INSTANCE_COUNT,
instance_type=estimator.train_instance_type,
data=inputs,
content_type="text/csv",
input_filter="$",
output_filter="$",
)
default_args = {
"owner": "airflow",
"start_date": airflow.utils.dates.days_ago(2),
"provide_context": True,
}
dag = DAG("tensorflow_example",
default_args=default_args,
schedule_interval="@once")
train_op = SageMakerTrainingOperator(task_id="tf_training",
config=training_config,
wait_for_completion=True,
dag=dag)
transform_op = SageMakerTransformOperator(task_id="transform_operator",
config=transform_config,
wait_for_completion=True,
dag=dag)
transform_op.set_upstream(train_op)
return training_config
def test_amazon_alg_model_config(sagemaker_session):
pca_model = pca.PCAModel(model_data="{{ model_data }}",
role="{{ role }}",
sagemaker_session=sagemaker_session)
config = airflow.model_config(instance_type="ml.c4.xlarge",
model=pca_model)
expected_config = {
"ModelName": "pca-%s" % TIME_STAMP,
"PrimaryContainer": {
"Image": "174872318107.dkr.ecr.us-west-2.amazonaws.com/pca:1",
"Environment": {},
"ModelDataUrl": "{{ model_data }}",
},
"ExecutionRoleArn": "{{ role }}",
}
assert config == expected_config
def test_amazon_alg_model_config(sagemaker_session):
pca_model = pca.PCAModel(model_data="{{ model_data }}",
role="{{ role }}",
sagemaker_session=sagemaker_session)
config = airflow.model_config(instance_type='ml.c4.xlarge',
model=pca_model)
expected_config = {
'ModelName': "pca-%s" % TIME_STAMP,
'PrimaryContainer': {
'Image': '174872318107.dkr.ecr.us-west-2.amazonaws.com/pca:1',
'Environment': {},
'ModelDataUrl': '{{ model_data }}'
},
'ExecutionRoleArn': '{{ role }}'
}
assert config == expected_config
def test_byo_framework_model_config(sagemaker_session):
byo_model = model.FrameworkModel(
model_data="{{ model_data }}",
image="{{ image }}",
role="{{ role }}",
entry_point="{{ entry_point }}",
source_dir="{{ source_dir }}",
env={"{{ key }}": "{{ value }}"},
name="model",
sagemaker_session=sagemaker_session,
)
config = airflow.model_config(instance_type="ml.c4.xlarge",
model=byo_model)
expected_config = {
"ModelName": "model",
"PrimaryContainer": {
"Image": "{{ image }}",
"Environment": {
"{{ key }}": "{{ value }}",
"SAGEMAKER_PROGRAM": "{{ entry_point }}",
"SAGEMAKER_SUBMIT_DIRECTORY":
"s3://output/model/source/sourcedir.tar.gz",
"SAGEMAKER_ENABLE_CLOUDWATCH_METRICS": "false",
"SAGEMAKER_CONTAINER_LOG_LEVEL": "20",
"SAGEMAKER_REGION": "us-west-2",
},
"ModelDataUrl": "{{ model_data }}",
},
"ExecutionRoleArn": "{{ role }}",
"S3Operations": {
"S3Upload": [{
"Path": "{{ source_dir }}",
"Bucket": "output",
"Key": "model/source/sourcedir.tar.gz",
"Tar": True,
}]
},
}
assert config == expected_config
def test_byo_model_config(sagemaker_session):
byo_model = model.Model(
model_data="{{ model_data }}",
image="{{ image }}",
role="{{ role }}",
env={"{{ key }}": "{{ value }}"},
name='model',
sagemaker_session=sagemaker_session)
config = airflow.model_config(instance_type='ml.c4.xlarge', model=byo_model)
expected_config = {
'ModelName': 'model',
'PrimaryContainer': {
'Image': '{{ image }}',
'Environment': {'{{ key }}': '{{ value }}'},
'ModelDataUrl': '{{ model_data }}'
},
'ExecutionRoleArn': '{{ role }}'
}
assert config == expected_config
def test_byo_framework_model_config(sagemaker_session):
byo_model = model.FrameworkModel(model_data="{{ model_data }}",
image="{{ image }}",
role="{{ role }}",
entry_point="{{ entry_point }}",
source_dir="{{ source_dir }}",
env={"{{ key }}": "{{ value }}"},
name='model',
sagemaker_session=sagemaker_session)
config = airflow.model_config(instance_type='ml.c4.xlarge',
model=byo_model)
expected_config = {
'ModelName': 'model',
'PrimaryContainer': {
'Image': '{{ image }}',
'Environment': {
'{{ key }}': '{{ value }}',
'SAGEMAKER_PROGRAM': '{{ entry_point }}',
'SAGEMAKER_SUBMIT_DIRECTORY':
's3://output/model/source/sourcedir.tar.gz',
'SAGEMAKER_ENABLE_CLOUDWATCH_METRICS': 'false',
'SAGEMAKER_CONTAINER_LOG_LEVEL': '20',
'SAGEMAKER_REGION': 'us-west-2'
},
'ModelDataUrl': '{{ model_data }}'
},
'ExecutionRoleArn': '{{ role }}',
'S3Operations': {
'S3Upload': [{
'Path': '{{ source_dir }}',
'Bucket': 'output',
'Key': 'model/source/sourcedir.tar.gz',
'Tar': True
}]
}
}
assert config == expected_config
def __init__(
self,
state_id,
model,
model_data_url=None,
sagemaker_submit_directory=None,
model_name=None,
instance_type=None,
tags=None,
**kwargs,
):
"""
Args:
state_id (str): State name whose length **must be** less than or
equal to 128 unicode characters. State names **must be** unique
within the scope of the whole state machine.
model (sagemaker.model.Model): The SageMaker model to use in the
ModelStep. If :py:class:`TrainingStep` was used to train the model
and saving the model is the next step in the workflow, the output
of :py:func:`TrainingStep.get_expected_model()` can be passed here.
model_name (str or Placeholder, optional): Specify a model name,
this is required for creating the model. We recommend to use
:py:class:`~stepfunctions.inputs.ExecutionInput` placeholder
collection to pass the value dynamically in each execution.
instance_type (str, optional): The EC2 instance type to deploy this
Model to. For example, 'ml.p2.xlarge'. This parameter is typically
required when the estimator used is not an `Amazon built-in
algorithm
<https://docs.aws.amazon.com/sagemaker/latest/dg/algos.html>`_.
tags (list[dict], optional): `List to tags
<https://docs.aws.amazon.com/sagemaker/latest/dg/API_Tag.html>`_ to
associate with the resource.
"""
if isinstance(model, FrameworkModel):
parameters = model_config(
model=model,
instance_type=instance_type,
role=model.role,
image=model.image,
)
if model_name:
parameters["ModelName"] = model_name
# placeholder for model data url
if model_data_url:
parameters["PrimaryContainer"]["ModelDataUrl"] = model_data_url
# placeholder for sagemaker script
if sagemaker_submit_directory:
parameters["PrimaryContainer"]["Environment"][
"SAGEMAKER_SUBMIT_DIRECTORY"
] = sagemaker_submit_directory
print(parameters)
elif isinstance(model, Model):
parameters = {
"ExecutionRoleArn": model.role,
"ModelName": model_name or model.name,
"PrimaryContainer": {
"Environment": {},
"Image": model.image,
"ModelDataUrl": model.model_data,
},
}
else:
raise ValueError(
(
f"Expected 'model' parameter to be of type 'sagemaker.model.Model'"
f", but received type '{type(model).__name__}'"
)
)
if "S3Operations" in parameters:
del parameters["S3Operations"]
if tags:
parameters["Tags"] = tags_dict_to_kv_list(tags)
kwargs[Field.Parameters.value] = parameters
kwargs[Field.Resource.value] = "arn:aws:states:::sagemaker:createModel"
super(MLMaxModelStep, self).__init__(state_id, **kwargs)
