A Comparison of Architectures and Pretraining Methods for Contextualized Multilingual Word Embeddings | Read Paper on Bytez