Multi-language is now available for bibliography annotation.
German, Portuguese and a mixed of french/english model has been pre-trained. A detection language step is done and fitted model is loaded on the fly.
Specific tokenization and lemmatization according to language is not yet implemented.