Closed amueller closed 1 year ago
I think we do here, no?https://github.com/automl/TabPFN/blob/889fad7070ded19ac3b247daf47d94b2538695cb/tabpfn/priors/flexible_categorical.py#L209
The idea is to remove datasets which have different subsets of classes in the train and test split.
indeed, you are right, my bad!
You're assigning
check_is_compatible
but inget_model
but then don't use it.