Accidentally passing sample weights with missing entries or for example 1 / 0 (as could happen if estimating inverse weights with kaplan-meier, which may yield estimates exactly = 0) went through previously. Adding a more stringent input check to prevent this - as this could be a rare occurrence when passing estimated weights to grf.
Accidentally passing sample weights with missing entries or for example 1 / 0 (as could happen if estimating inverse weights with kaplan-meier, which may yield estimates exactly = 0) went through previously. Adding a more stringent input check to prevent this - as this could be a rare occurrence when passing estimated weights to grf.