跳转至

Python–张量流。devicespec.parse_from_string()

原文:https://www.geesforgeks.org/python-tensorflow-device spec-parse_from_string/

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。

parse_from_string() 用于将设备规格名称解析为其组件。

语法:张量流。device spec.parse_from_string(spec)

参数:

  • 规格:是格式/作业:/副本:/任务:/设备:CPU:或/作业:/副本:/任务:/设备:GPU 的字符串,所有字段都是可选的。

返回:返回一个 DeviceSpec 对象。

例 1:

蟒蛇 3

# Importing the library
import tensorflow as tf

# Initializing Device Specification
device_spec = tf.DeviceSpec(job ="gfg", replica = 5)

# Printing the DeviceSpec 
print('Device Spec: ', device_spec.to_string())

# Getting new DeviceSpec object
new_device_spec = device_spec.parse_from_string("/GPU:0")

# Printing the result
print('New Device Spec: ', new_device_spec.to_string())

输出:

Device Spec:  /job:gfg/replica:5
New Device Spec:  /device:GPU:0

例 2:

蟒蛇 3

# Importing the library
import tensorflow as tf

# Initializing Device Specification
device_spec = tf.DeviceSpec(job ="gfg", replica = 5)

# Printing the DeviceSpec 
print('Device Spec: ', device_spec.to_string())

# Getting new DeviceSpec object
new_device_spec = device_spec.parse_from_string("replica:2 / GPU:0")

# Printing the result
print('New Device Spec: ', new_device_spec.to_string())

输出:

Device Spec:  /job:gfg/replica:5
New Device Spec:  /replica:2/device:GPU:0



回到顶部