Large Language Models Can Be Strong Differentially Private Learners.

Xuechen Li, Florian Tramr, Percy Liang, Tatsunori Hashimoto

Browse the full ICLR paper archive.