Have you ever tried to generate rectangular images?

[pupuchuan]

The resolution of my dataset images is 256×512, and I want the initial feature map F0 to have size 2×4×512. I tried to changed the p size into 8×512 while keeping the size of z at 16×512. However the attention module does not seem to train properly in this case. So, do I have to compress the dimensions of z to be the same as p since I don't want to change the resolution of the input image? Or is there any solution for this problem?

Thank you!

[BillyXYB]

Thank you for your question
We have not tried to generate rectangular images.
The dimension of Z and P should be identical for Attention. Maybe you could also have an initial P(16512) for interaction, and transform it into F0(8512).