def __init__()

in models/src/wavegrad/downsampling.py [0:0]

• 38 lines of code
• 4 McCabe index (conditional complexity)

    def __init__(self, in_channels, out_channels, factor, dilations):
        super(DownsamplingBlock, self).__init__()
        in_sizes = [in_channels] + [out_channels for _ in range(len(dilations) - 1)]
        out_sizes = [out_channels for _ in range(len(in_sizes))]
        self.main_branch = torch.nn.Sequential(
            *(
                [
                    InterpolationBlock(
                        scale_factor=factor,
                        mode="linear",
                        align_corners=False,
                        downsample=True,
                    )
                ]
                + [
                    ConvolutionBlock(in_size, out_size, dilation)
                    for in_size, out_size, dilation in zip(
                        in_sizes, out_sizes, dilations
                    )
                ]
            )
        )
        self.residual_branch = torch.nn.Sequential(
            *[
                Conv1dWithInitialization(
                    in_channels=in_channels,
                    out_channels=out_channels,
                    kernel_size=1,
                    stride=1,
                ),
                InterpolationBlock(
                    scale_factor=factor,
                    mode="linear",
                    align_corners=False,
                    downsample=True,
                ),
            ]
        )