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:
Python 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:
Python 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)
版权属于：月萌API www.moonapi.com，转载请注明出处
