A novel unified model called UNIMASK-M that can effectively address various human motion synthesis tasks using a single architecture, achieving comparable or better performance than state-of-the-art task-specific models.
