Hello. First, congratulations for this work. Could you please explain why the inconsistency weight has been set to zero? Does the latest configurations present better performance than the ones related in the paper? Thanks for your attention.