def remove_weightnorm(self):
for wavenet in self.wavenet:
wavenet.start = torch.nn.utils.remove_weight_norm(wavenet.start)
wavenet.in_layers = remove(wavenet.in_layers)
wavenet.cond_layer = torch.nn.utils.remove_weight_norm(
wavenet.cond_layer)
wavenet.res_skip_layers = remove(wavenet.res_skip_layers)
def remove_weightnorm(model):
squeezewave = model
for wavenet in squeezewave.wavenet:
wavenet.start = torch.nn.utils.remove_weight_norm(wavenet.start)
wavenet.in_layers = remove_batchnorm(wavenet.in_layers)
wavenet.cond_layer = torch.nn.utils.remove_weight_norm(
wavenet.cond_layer)
wavenet.res_skip_layers = remove(wavenet.res_skip_layers)
return squeezewave
def remove_weightnorm(self):
self.wn.start = torch.nn.utils.remove_weight_norm(self.wn.start)
self.wn.in_layers = remove(self.wn.in_layers)
self.wn.cond_layer = torch.nn.utils.remove_weight_norm(
self.wn.cond_layer)
self.wn.res_skip_layers = remove(self.wn.res_skip_layers)
