Models that can, given information about an ongoing dialogue, can produce a label that can be given to a GRED model that generate a specific emotion.