pytorch之torch.flip

torch.flip(input,dim):第一个参数是输入，第二个参数是输入的第几维度，按照维度对输入进行翻转

import torch
x = torch.arange(16).view(2, 2, 2,2)
print('x=\n',x)
a = torch.flip(x, [2])
print('a=\n',a)

x=
 tensor([[[[ 0,  1],
          [ 2,  3]],

         [[ 4,  5],
          [ 6,  7]]],


        [[[ 8,  9],
          [10, 11]],

         [[12, 13],
          [14, 15]]]])
a=
 tensor([[[[ 2,  3],
          [ 0,  1]],

         [[ 6,  7],
          [ 4,  5]]],


        [[[10, 11],
          [ 8,  9]],

         [[14, 15],
          [12, 13]]]])

Process finished with exit code 0

import torch
x = torch.arange(16).view(2, 2, 2,2)
print('x=\n',x)
a = torch.flip(x, [3])
print('a=\n',a)

x=
 tensor([[[[ 0,  1],
          [ 2,  3]],

         [[ 4,  5],
          [ 6,  7]]],


        [[[ 8,  9],
          [10, 11]],

         [[12, 13],
          [14, 15]]]])
a=
 tensor([[[[ 1,  0],
          [ 3,  2]],

         [[ 5,  4],
          [ 7,  6]]],


        [[[ 9,  8],
          [11, 10]],

         [[13, 12],
          [15, 14]]]])

Process finished with exit code 0