We currently use bert-as-service for sentence encoding/embedding by mapping a sentence to a fixed length vector using BERT models. Are there alternative NLP libraries which does