I'd be very surprised if this is in the training data given that most models mess it up to this day. E.g. look at the ones from Opus.