How to use with other languages other than english?

I would like to use KeyBert with the French language. To do this, must I select model and pass it through KeyBERT with model? Like this:

from keybert import KeyBERT
doc = """
L'apprentissage supervisé est la tâche d'apprentissage machine qui consiste à apprendre une fonction qui associe une entrée à une sortie en se basant sur des exemples de paires entrée-sortie [1]. 
Il déduit une fonction à partir de données de formation étiquetées consistant en un ensemble d'exemples de formation.
Dans l'apprentissage supervisé, chaque exemple est une paire constituée d'un objet d'entrée  (généralement un vecteur) et une valeur de sortie souhaitée (également appelée signal de supervision). 
"""
model = KeyBERT(model='MODEL_TO_CHOOSE')
keywords = model.extract_keywords(doc)

For the French language:

• Which model do you recommend?
• Is that xlm-r-bert-base-nli-stsb-mean-tokens is a good choice?

[MaartenGr]

Hmmm, it might improve if you increase the keyphrase_ngram_range to (1, 3) for example. However, this is exactly what can happen with KeyBERT. It is highly dependent on the underlying embedding model. For that reason, I typically like to use embedding models that are trained mostly for similarity measures, like those used in sentence-transformers (e.g., xlm-r-bert-base-nli-stsb-mean-tokens).

All in all, it requires some experimentation with models until you find the one best suited to your use-case.

[LeMoussel]

Compare keyword extraction results, in French language, from TF/IDF, Yake, KeyBert: https://gist.github.com/LeMoussel/aa5f7c46c6cb09473e97b20eb2e13cc4 The quality of keywords with KeyBert seems “strange” to me. 🤔