BlendCSE: Blend contrastive learnings for sentence embeddings with rich semantics and transferability

Jiahao Xu, Charlie Soh Zhanyi, Liwen Xu, Lihui Chen

Abstract: Highlights•Sentence representation for rich semantics and strong transferability.•Sentence representation via contrastive learning compatible with data augmentation.•Blends supervised, self-supervised learning and masked language modeling.•Achieves state-of-the-art for downstream tasks and document applications.