Currently, it is incumbent on the user to know which pipeline configuration they used when creating a model. If one loads a model that they had saved, and do not use exactly the same pipeline configuration, the program will most likely fail, though there might be a few coincidental cases where it works or silently performs unexpected behavior. One could currently get one of several uninformative errors.
We should seek to validate that the model and the pipeline are compatible when the model is loaded and raise an exception that conveys as much when this is not the case.
Currently, it is incumbent on the user to know which pipeline configuration they used when creating a model. If one loads a model that they had saved, and do not use exactly the same pipeline configuration, the program will most likely fail, though there might be a few coincidental cases where it works or silently performs unexpected behavior. One could currently get one of several uninformative errors.
We should seek to validate that the model and the pipeline are compatible when the model is loaded and raise an exception that conveys as much when this is not the case.