I can add that if I convert the data type in spark

if I use toPandas() --> then it works

if I use to_pandas_on_spark() -->same error