Hi! Creator of the library here. If you change the generation parameters to be greedy (i.e., sample=no and top_k=0), you will get "Bonjour, comment êtes-vous?"
The top_k and sample generation parameters are just there to show that they are supported :), and is sometimes useful for the other tasks (like text generation w/gpt2, to get more variety)
I understand there's reasons the translation is incorrect, but if the very first example you're showing on the page is wrong, most people (who are fluent enough) will just roll their eyes and leave it at that. Maybe showcase an example that works?
Which the model likes to translate to,
Which no french person would ever say to you because that's a lot of words and doesn't really sound very... French.And of course are you talking to someone familiar... so on and so forth.