transpose在numpy和torch中的不同

numpy和torch中transpose的功能不同

numpy.ndarray.transpose的官方文档
torch.transpose的官方文档

numpy.transpose需指定一个新的axis的顺序

import numpy as np
array_ = np.arange(24).reshape(1,2,3,4)
print(array_)
# [[[[ 0  1  2  3]
#    [ 4  5  6  7]
#    [ 8  9 10 11]]
# 
#   [[12 13 14 15]
#    [16 17 18 19]
#    [20 21 22 23]]]]

array_t = array_.transpose([0,3,2,1])
print(array_t)
# [[[[ 0 12]
#    [ 4 16]
#    [ 8 20]]
# 
#   [[ 1 13]
#    [ 5 17]
#    [ 9 21]]
# 
#   [[ 2 14]
#    [ 6 18]
#    [10 22]]
# 
#   [[ 3 15]
#    [ 7 19]
#    [11 23]]]]

array_tt = array_.transpose([0,3,2,1])
print(array_tt)