tensorflow:检验张量秩

Question

我想检查张量的等级。下面是我的代码：

import tensorflow as tf
x = tf.constant([[0,1,0], [0,1,0]])
print(tf.rank(x))

它回来了

Tensor("Rank_14:0", shape=(), dtype=int32)

Rank_14:0不断增长的地方。我希望它还能回来。我做错什么了？

Answer 1

Rank_14:0是返回的张量的名称，而不是它的值，您需要在会话中计算张量以获得实际值：

with tf.Session() as sess:
    sess.run(tf.rank(x))

import tensorflow as tf
x = tf.constant([[0,1,0], [0,1,0]])
r = tf.rank(x)

sess = tf.InteractiveSession()
​
print("My tensor is: ", r)
print("The value of my tensor is: ", r.eval())

My tensor is:  Tensor("Rank_3:0", shape=(), dtype=int32)
The value of my tensor is:  2