Could I just use this model for type classification? Because I see that a lot of work should first identify repeat et al.