External_attention in the code is strange, only do attention on channel, but no attention on pixel.

[zanonShao]

Hello Menhao,
I have some question about the difference between code and paper.
In the paper, 'Equation (5) is the similarity between the i-th pixel and the j-th rows of M', external attention do the attention between pixels, however in the code I think Conv1d only can do the attention amount only one pixels’ channels, and pixels in F not do attention with pixels in M.
And in the paper 'In fact, we find that a small _S_, e.g. 64, works well in experiments.' but in the code d is setted to 64 instead of S.

[YouJiacheng]

M is a set of "pattern" feature vectors (#vector = S), not feature vectors of pixels.
Conv1d.weight can be viewed as M, #out channel = S, # in channel = d.
Therefore the result of Conv1d will be the dot product between feature vectors of pixels and M's rows(shape of M = S × d = out × in, out channels ↔ rows).
k in code = S in paper, c in code = d in paper.
等一下，为什么你要用英文提问，作者也是中国人啊。

[zanonShao]

ok, I get it.
谢谢你还先用英文回答🤣