espnet2.gan_codec.shared.discriminator.msmpmb_discriminator.WNConv1d

source

espnet2.gan_codec.shared.discriminator.msmpmb_discriminator.WNConv1d(*args, **kwargs)

Create a 1D convolutional layer with weight normalization.

This function wraps the standard nn.Conv1d layer with weight normalization. Optionally, it can also include a Leaky ReLU activation layer.

• Parameters:
  • *args – Variable length argument list for the convolution layer.
  • **kwargs – Keyword arguments for the convolution layer, including: act (bool): If True, adds a LeakyReLU activation. Default is True.
• Returns: The convolutional layer, possibly wrapped in a Sequential model with activation.
• Return type: nn.Sequential or nn.Conv1d

Examples

>>> conv_layer = WNConv1d(1, 16, kernel_size=3)
>>> print(conv_layer)
Sequential(
  (0): Conv1d(1, 16, kernel_size=(3,), stride=(1,), ...)
  (1): LeakyReLU(negative_slope=0.1)
)