The Hive UDF won't help you. If you look at the Hive issue tracker https://issues.apache.org/jira/browse/HIVE-1304, row_sequence() was added as a workaround because they didn't support the row_number() analytic function at that point in time.

We support the row_number() analytic function in Impala, so there's no reason to try to use that UDF.

If you want to start froma particular number, can't you just add it like I suggested in my previous answer?