Is pyspark.sql grouped data putting each group into a single partition?

Question

When I use group aggregate pandas UDFs in GroupedData.agg(), how will this UDF be applied to the data in a single group? I would think that spark is first partitioning the data