You will want to use a two channel conv1d as the first convolution later. I.e. it will take in a tensor of shape [B, 2, 18].
確定! 回上一頁