gluonts.nursery.sagemaker_sdk.model module

class gluonts.nursery.sagemaker_sdk.model.GluonTSModel(model_data, role, entry_point, image: str = None, framework_version: str = '0.4.1', predictor_cls=<class 'gluonts.nursery.sagemaker_sdk.model.GluonTSPredictor'>, model_server_workers: int = None, **kwargs)[source]

Bases: sagemaker.model.FrameworkModel

An GluonTS SageMaker Model that can be deployed to a SageMaker Endpoint.

prepare_container_def(instance_type, accelerator_type=None) → Dict[str, str][source]

Return a container definition with framework configuration set in model environment variables.

Parameters
  • instance_type

    The EC2 instance type to deploy this Model to. Example:

    'ml.c5.xlarge' # CPU,
    'ml.p2.xlarge' # GPU.
    

  • accelerator_type

    The Elastic Inference accelerator type to deploy to the instance for loading and making inferences to the model. Example:

    "ml.eia1.medium"
    

Returns

A container definition object usable with the CreateModel API.

Return type

Dict[str, str]

class gluonts.nursery.sagemaker_sdk.model.GluonTSPredictor(endpoint_name: str, sagemaker_session: sagemaker.session.Session = None)[source]

Bases: sagemaker.predictor.RealTimePredictor

A RealTimePredictor for inference against GluonTS Endpoints. This is able to serialize and deserialize datasets in the gluonts data format.