Yield also won’t work for a billions rows datasets, and pandas reads can be chunked and processed much more efficiently than generators