IS there a deeper reason these objects get cloned? The information of the models is quite valuable. I mean, the information seems to be somewhere, since "predict" obviously uses the fitted model 🙂 . Is there an easy way to access the specific (and fitted) "model" object inside the MLForecast class?