The output feature tensor from the _forward_features method will have shape (batch_size, n, n, channels). We can reshape it(flatten it) using view(x.size[0], -1) ...
確定! 回上一頁