tensorflow 打印内存中的变量方法

时间:2022-11-19 16:36:00

法一:

循环打印

模板

?
1
2
for (x, y) in zip(tf.global_variables(), sess.run(tf.global_variables())):
 print '\n', x, y

实例

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
# coding=utf-8
 
import tensorflow as tf
 
 
def func(in_put, layer_name, is_training=True):
 with tf.variable_scope(layer_name, reuse=tf.AUTO_REUSE):
  bn = tf.contrib.layers.batch_norm(inputs=in_put,
           decay=0.9,
           is_training=is_training,
           updates_collections=None)
 return bn
 
def main():
 
 with tf.Graph().as_default():
  # input_x
  input_x = tf.placeholder(dtype=tf.float32, shape=[1, 4, 4, 1])
  import numpy as np
  i_p = np.random.uniform(low=0, high=255, size=[1, 4, 4, 1])
  # outputs
  output = func(input_x, 'my', is_training=True)
  with tf.Session() as sess:
   sess.run(tf.global_variables_initializer())
   t = sess.run(output, feed_dict={input_x:i_p})
 
   # 法一: 循环打印
   for (x, y) in zip(tf.global_variables(), sess.run(tf.global_variables())):
    print '\n', x, y
 
if __name__ == "__main__":
 main()
?
1
2
3
4
5
6
7
8
9
2017-09-29 10:10:22.714213: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1052] Creating TensorFlow device (/device:GPU:0) -> (device: 0, name: GeForce GTX 1070, pci bus id: 0000:01:00.0, compute capability: 6.1)
 
<tf.Variable 'my/BatchNorm/beta:0' shape=(1,) dtype=float32_ref> [ 0.]
 
<tf.Variable 'my/BatchNorm/moving_mean:0' shape=(1,) dtype=float32_ref> [ 13.46412563]
 
<tf.Variable 'my/BatchNorm/moving_variance:0' shape=(1,) dtype=float32_ref> [ 452.62246704]
 
Process finished with exit code 0

法二:

指定变量名打印

模板

?
1
print 'my/BatchNorm/beta:0', (sess.run('my/BatchNorm/beta:0'))

实例

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
# coding=utf-8
 
import tensorflow as tf
 
 
def func(in_put, layer_name, is_training=True):
 with tf.variable_scope(layer_name, reuse=tf.AUTO_REUSE):
  bn = tf.contrib.layers.batch_norm(inputs=in_put,
           decay=0.9,
           is_training=is_training,
           updates_collections=None)
 return bn
 
def main():
 
 with tf.Graph().as_default():
  # input_x
  input_x = tf.placeholder(dtype=tf.float32, shape=[1, 4, 4, 1])
  import numpy as np
  i_p = np.random.uniform(low=0, high=255, size=[1, 4, 4, 1])
  # outputs
  output = func(input_x, 'my', is_training=True)
  with tf.Session() as sess:
   sess.run(tf.global_variables_initializer())
   t = sess.run(output, feed_dict={input_x:i_p})
 
   # 法二: 指定变量名打印
   print 'my/BatchNorm/beta:0', (sess.run('my/BatchNorm/beta:0'))
   print 'my/BatchNorm/moving_mean:0', (sess.run('my/BatchNorm/moving_mean:0'))
   print 'my/BatchNorm/moving_variance:0', (sess.run('my/BatchNorm/moving_variance:0'))
 
if __name__ == "__main__":
 main()
?
1
2
3
4
5
6
7
2017-09-29 10:12:41.374055: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1052] Creating TensorFlow device (/device:GPU:0) -> (device: 0, name: GeForce GTX 1070, pci bus id: 0000:01:00.0, compute capability: 6.1)
 
my/BatchNorm/beta:0 [ 0.]
my/BatchNorm/moving_mean:0 [ 8.08649635]
my/BatchNorm/moving_variance:0 [ 368.03442383]
 
Process finished with exit code 0

以上这篇tensorflow 打印内存中的变量方法就是小编分享给大家的全部内容了,希望能给大家一个参考,也希望大家多多支持服务器之家。

原文链接:https://blog.csdn.net/JNingWei/article/details/78131214