Python–张量流。gradienttape.stop_recording()

原文:https://www.geesforgeks.org/python-tensorflow-gradienttape-stop_recording/

TensorFlow 是谷歌设计的开源 Python 库，用于开发机器学习模型和深度学习神经网络。

stop_recording() 用于暂时停止记录操作。如果磁带没有录制，将会出现错误。

语法:张量流。GradientTape.stop_recording()

参数:不接受任何参数。

返回:无

加注:

• 运行时间错误:如果磁带没有正确记录，它将引发运行时间错误。

例 1:

蟒蛇 3

# Importing the library
import tensorflow as tf

x = tf.constant(4.0)

# Using GradientTape
with tf.GradientTape() as gfg:
  gfg.watch(x)

  # Stop recording
  with gfg.stop_recording():
    y = x * x * x

# Computing gradient
res = gfg.gradient(y, x) 

# Printing result
print("res: ", res)

输出:

res:  None

例 2:

蟒蛇 3

# Importing the library
import tensorflow as tf

x = tf.constant(4.0)

# Using GradientTape
with tf.GradientTape() as gfg:
  gfg.watch(x)

  # Stop recording
  with gfg.stop_recording():
    y = x * x * x

  # Starting the recording again
  gfg.watch(x)
  y = x * x

# Computing gradient
res = gfg.gradient(y, x) 

# Printing result
print("res: ", res)

输出:

res:  tf.Tensor(8.0, shape=(), dtype=float32)

---

我们一直在努力

apachecn/AiLearning

---