When given an incorrect file/directory path as a data source, the resultant error says: 'Loading from premade dataset not implemented yet'.
This seems to be because the code assumes that if the path doesn't exist, we must be trying to load a pre-made dataset, but this isn't always the case.