French lemmatizer lemmatizes differently Expliquez and expliquez

[alephpi]

Hi, I'm new to spaCy and I'm currently using fr_core_news_sm model to do some NLP job. I find that the lemmatizer (when calling Token.lemma_) doesn't treat Expliquez and expliquez evenly, where the first lemma is expliquez and the second lemma is expliquer (what I want indeed). I wonder

1. why the lemmatizer has this discrimination?
2. what is the best practice to get the truly lemma (whose lemma is itself)? A quick and dirty way is of course to do nlp(lemma) repeatedly but is there a better practice?

[alephpi]

More observations:
expliquez is falsely pos-tagged as ADP rather than VERB

[alephpi]

Well, turns out that better use the large model fr_core_news_lg, that way all the issues are gone.

[svlandeg]

Thanks for reporting back! In general, the large ("lg") models will indeed be more accurate than the small ("sm") ones.