about pixel-cnn/model.py, line38

[Lucas2012]

Hi, thanks for the code!

I have a question about pixel-cnn/model.py, line38. It's appending ones to the tensor x. But I'm wondering how this is used -- it seems that after the "convolution(x_pad)" the ones channel is already disappeared. And I didn't find the corresponding code for using the ones channel to distinguish the image out. And more is that since pixelcnn is doing shifting correction for every time, it seems that there's no need for further distinguishing image x.

[share020]

I'm not so clear about that too. Have you figured it out?

[saehoonkim]

I guess that we don't need to append a channel to the input tensor x anymore. I just obtained the similar bpd even after removing this appending step in a pytorch version.