Parameters of Upblock

[reubengann]

I am trying to reproduce this work. In The definition of DeconvBlock, your parameters are:

class DeconvBlock(torch.nn.Module):
    def __init__(self, input_size, output_size, kernel_size=4, stride=2, padding=1, bias=True, activation='prelu', norm=None):

When you go to use this class in, for instance, Upblock, you pass arguments:

class UpBlock(torch.nn.Module):
    def __init__(self, num_filter, kernel_size=8, stride=4, padding=2, bias=True, activation='prelu', norm=None):
        super(UpBlock, self).__init__()
        self.up_conv1 = DeconvBlock(num_filter, num_filter, kernel_size, stride, padding, activation, norm=None)

I don't understand how this can work. The 6th parameter of DeconvBlock is supposed to be bias (bool) but here your 6th argument is activation (str). How does this code work?