arrays - 数组的转换产生一个空占位符

Question

我有一个名为 new_input_processed 的 numpy 数组。下面的代码将其转换为 float32 类型的单热数组(参见 byte_list)。但是当我输入 byte_list 来查看该数组的值时，我得到一个空张量。我想要一个非空张量。是否可以 ？

In [30]: new_input_processed
Out[30]: 
array([[ 83, 111, 109, 101,  32,  83, 101, 113, 117, 101, 110,  99, 101,
         32, 111, 102,  32,  99, 104,  97, 114,  97,  99, 116, 101, 114,
          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
          0,   0,   0,   0,   0,   0,   0,   0,   0]], dtype=uint8)

In [31]: byte_list = tf.cast(tf.one_hot(new_input_processed, 256, 1, 0), dtype=tf.float32)

In [32]: byte_list
Out[32]: <tf.Tensor 'Cast_2:0' shape=(1, 100, 256) dtype=float32>

Answer 1

您没有得到空张量。张量对象信息正确返回:

<tf.Tensor 'Cast_2:0' shape=(1, 100, 256) dtype=float32>

看形状，和想象的一样。

尽管如此，如果您想查看内容(即 byte_list Tensor 对象的实际值)，一种方法是调用 eval() .

类似这样的事情应该做:

import numpy as np
import tensorflow as tf
new_input_processed = np.array([[ 83, 111, 109, 101,  32,  83, 101, 113, 117, 101, 110,  99, 101,
                                  32, 111, 102,  32,  99, 104,  97, 114,  97,  99, 116, 101, 114,
                                  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
                                  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
                                  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
                                  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
                                  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,
                                  0,   0,   0,   0,   0,   0,   0,   0,   0]], dtype=np.uint8)

byte_list = tf.cast(tf.one_hot(new_input_processed, 256, 1, 0), dtype=tf.float32)

with tf.Session() as sess:  print(byte_list.eval())  # here

输出:

[[[0. 0. 0. ... 0. 0. 0.]
  [0. 0. 0. ... 0. 0. 0.]
  [0. 0. 0. ... 0. 0. 0.]
  ...
  [1. 0. 0. ... 0. 0. 0.]
  [1. 0. 0. ... 0. 0. 0.]
  [1. 0. 0. ... 0. 0. 0.]]]