about A = -torch.exp(self.A_log.float())

why add "-" in this code?why not use A directly instead of A_log? A = -torch.exp(self.A_log.float())