I am working with spacy on french text, but it seems to me that there is some consistency issue :
nlp = spacy.load(\'fr_core_news_md\') print([i.lemma_ for i in n
