Not so sure about the patch embedding kernel size.

[Asagami-Fujino]

Thanks a lot for your work，it helps a lot.
In your paper you mentioned that the convolution kernel size is a conbination of 1x14x14 and 14x1x1, but in the code it seems that the kernel size is 1x16x16 and 16x1x1. So which is correct?

[peterant330]

Thanks a lot for your work，it helps a lot. In your paper you mentioned that the convolution kernel size is a conbination of 1x14x14 and 14x1x1, but in the code it seems that the kernel size is 1x16x16 and 16x1x1. So which is correct?

Hi,

It should be 16. There is a typo in the paper. Thanks for pointing this out.