The input is padded, not the output. In your case, the conv2d layer will apply a two-pixel padding on all sides just before computing the ...
確定! 回上一頁