UOR: Universal Backdoor Attacks on Pre-trained Language Models.

Wei Du, Peixuan Li, Haodong Zhao, Tianjie Ju, Ge Ren, Gongshen Liu

Browse the full ACL paper archive.