[Python] Is predict_proba function thread safe, when callin model.predict_proba from multiple...

I have some_ai_model.sav model that I want to integrate using python FastAPI. I'm wondering: is the model’s predict_proba function thread-safe? If multiple requests come into the app, would it be a problem if I only have a single instance of the model loaded using joblib? Or should I create a model pool? What do you recommend?

Continue reading...