博客

tf.transpose(
    a,
    perm=None,
    name='transpose',
    conjugate=False
)

作用：转置、共轭转置等。

参数

perm

： returned tensor’s dimension

i

will correspond to the input dimension

perm[i]


例子：

test = tf.random_normal([2,2,3],dtype=tf.float32)
with tf.Session() as sess:
    #默认axis=0
    #print(sess.run([test,tf.transpose(test)]),"\n")
    print(sess.run([test,tf.transpose(test,perm=[1,2,0])]),"\n")
    #print(sess.run([test,tf.transpose(test,perm=[1, 0, 2])]))



输出：
[array([[[ 0.8021455 ,  0.505679  ,  1.8120127 ],
        [-0.03189123,  1.1534904 ,  0.9760273 ]],

       [[ 1.2107425 , -0.66847354,  0.73315024],
        [ 1.304936  ,  1.5720342 ,  0.6012646 ]]], dtype=float32), array([[[ 0.8021455 ,  1.2107425 ],
        [ 0.505679  , -0.66847354],
        [ 1.8120127 ,  0.73315024]],

       [[-0.03189123,  1.304936  ],
        [ 1.1534904 ,  1.5720342 ],
        [ 0.9760273 ,  0.6012646 ]]], dtype=float32)] 

相当于对矩阵为元素的numpy array做

转置

。

二维的情况：

x = tf.constant([[1, 2, 3], [4, 5, 6]])
tf.transpose(x)  # [[1, 4]
                 #  [2, 5]
                 #  [3, 6]]

# Equivalently
tf.transpose(x, perm=[1, 0])  # [[1, 4]
                              #  [2, 5]
                              #  [3, 6]]