1 paper
M. Li, H. Bai, L. Tan +2
Measuring sentence semantic similarity using pre-trained language models such as BERT generally yields unsatisfactory zero-shot performance, and one main reason is ineffective toke…