A collection of embedding models for sentence similarity, reranking and fill-mask tasks