This is your work, valued
SpanishTransformerXL. Language model trained on wiki corpus (500M tokens) with fastai v1 acc>42.3% len(vocab)=60K
