components batch_resource_manager

Batch Inference Resource Manager

batch_resource_manager

Overview

Resource Manager for batch inference.

Version: 0.0.1

View in Studio: https://ml.azure.com/registries/azureml/components/batch_resource_manager/version/0.0.1

Inputs

Name	Description	Type	Default	Optional	Enum
wait_input		uri_file		True
endpoint_name	The endpoint name or url. If not provided, the endpoint will be created.	string		True
deployment_name	The deployment name. If not provided, the deployment will be created.	string		True
model	Model name or model asset path. Only needed for managed deployment.	string		True
model_version	Model version. If model asset path provided, this one can be optional.	string		True
deployment_sku	The sku for the deployment. Only needed for managed deployment.	string		True
endpoint_workspace	The workspace name for the endpoint. If not provided, the same workspace of the run will be used.	string		True
endpoint_resource_group	The resource group name for the endpoint. If not provided, the same resource group of the run will be used.	string		True
endpoint_subscription_id	The subscription id for the endpoint. If not provided, the same subscription id of the run will be used.	string		True
endpoint_location	The location for the endpoint. Only required for the AOAI endpoint. If not provided, the same location of the run will be used.	string		True
connections_name	The connections name.	string		True
delete_managed_resources	If true, the managed resources created during the run will be deleted.	boolean	True	True
deployment_metadata_dir	The deployment metadata directory that contains deployment details.	uri_folder		True

Outputs

Name	Description	Type
output_metadata	Path to the folder where the deployment metadata will be stored.	uri_folder

Environment

azureml://registries/azureml/environments/model-evaluation/labels/latest