The sample function in the vignette needs to know how many classes were in the training data. We could pretty easily supply that to any fn that has an argument with some specific name indicating they want it. Maybe put some other basic training-set summary info into the "model" as well to be referenced via named arguments? That gets dangerously close to actually TRAINING the model-function, but it could unlock some very interesting options!
The sample function in the vignette needs to know how many classes were in the training data. We could pretty easily supply that to any
fnthat has an argument with some specific name indicating they want it. Maybe put some other basic training-set summary info into the "model" as well to be referenced via named arguments? That gets dangerously close to actually TRAINING the model-function, but it could unlock some very interesting options!