在应用其上方池层之后，如何打印张量对象的值

Question

将其应用于Tensorflow变量后，我只能获得对象的形状，如何在这些层之后获取变量的值？

def model_part2(a):  
    #q=tf.global_variables_initializer()
    p=tf.keras.layers.Conv1D(192,1)(a)
    #print(p.eval())
    p=tf.keras.layers.ReLU()(p)
    p=tf.keras.layers.MaxPool1D(1,2)(p)
    p=tf.keras.layers.Conv1D(256,1)(p)
    p=tf.keras.layers.ReLU()(p)
    p=tf.keras.layers.MaxPool1D(1,2)(p)
    p=tf.keras.layers.Conv1D(512,1)(p)
    p=tf.keras.layers.ReLU()(p)
    p=tf.keras.layers.MaxPool1D(1,2)(p)
    return p

`

Answer 1

在 Tensorflow Version 2.x 中，它非常简单。工作代码如下：

import tensorflow as tf
a =tf.constant(1.0,shape=[128,8192,1])
def model_part2(a):  
    #q=tf.global_variables_initializer()
    p=tf.keras.layers.Conv1D(192,1)(a)
    #print(p.eval())
    p=tf.keras.layers.ReLU()(p)
    p=tf.keras.layers.MaxPool1D(1,2)(p)
    p=tf.keras.layers.Conv1D(256,1)(p)
    p=tf.keras.layers.ReLU()(p)
    p=tf.keras.layers.MaxPool1D(1,2)(p)
    p=tf.keras.layers.Conv1D(512,1)(p)
    p=tf.keras.layers.ReLU()(p)
    p=tf.keras.layers.MaxPool1D(1,2)(p)
    return p
q = model_part2(a)
print(q)

以上代码的输出如下：

tf.Tensor(
[[[0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  ...
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]]
 [[0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  ...
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]]
 [[0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  ...
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]]
 ...
 [[0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  ...
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]]
 [[0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  ...
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]]
 [[0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  ...
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]
  [0.06378555 0.00853285 0.03427356 ... 0.         0.         0.        ]]], shape=(128, 1024, 512), dtype=float32)