why some parameter is different between paper and code?

[kingstarcraft]

In the paper: Our pyramidal decomposition architecture is illustrated in Figure 3. In practice, we extract across a total of 6 scales. The feature extractors for the individual scales are composed of a sequence of convolutions with kernels or and ReLUs with a leak of 0.2. We learn all downsamplers as convolutions with a stride of 2. But from the code, i found the kernel size of downsample is 3

[sftwre]

That is a mistake, thanks for pointing that out.