Learn FastText-style word representations from tokenized documents and compose embeddings from character n-grams.
