You have an Azure Machine Learning workspace. You are using the uri_folder data type as the output for a pipeline component. Which data access mode should you define that is supported by this configuration?