BPEmb: Tokenization-free Pre-trained Subword Embeddings in 275 Languages

Publication
Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)