Compressing Neural Machine Translation Models with 4-bit Precision

ACL 2020