created DataSet ArgType representing the common API for array and graph. Through this dataset, adapters can specify their input preferences between 'array' and 'graph' backends, and the pipeline chooses the best one. There are two properties as input to DataSet ArgType:
preference - This only applies to input datasets and is ignored for output datasets. It accepts values "array" or "graph" representing the 2 backends of the API, or None if no preference.
input_ref - This only applies to output datasets and is ignored for input datasets. It accepts a string value which can refer to any dataset input to the same adapter. This is used to indicate which input dataset the adapter has modified and sending as output. If the output dataset is new and not modified from any input, then the value can be None (default)
created DataSet ArgType representing the common API for array and graph. Through this dataset, adapters can specify their input preferences between 'array' and 'graph' backends, and the pipeline chooses the best one. There are two properties as input to DataSet ArgType:
preference - This only applies to input datasets and is ignored for output datasets. It accepts values "array" or "graph" representing the 2 backends of the API, or None if no preference.
input_ref - This only applies to output datasets and is ignored for input datasets. It accepts a string value which can refer to any dataset input to the same adapter. This is used to indicate which input dataset the adapter has modified and sending as output. If the output dataset is new and not modified from any input, then the value can be None (default)