One size does not fit all: Investigating strategies for differentially-private learning across NLP tasks

Manuel Senge, Timour Igamberdiev, Ivan Habernal