Dear authors, Have you released your pretrained model for the hybrid training paradiagm? If yes, could you send me a link please?