1. Creating a virtual environment

conda create -n tf1.14 python=3.7

2. Activate the environment

conda activate tf1.14

3. Install the specified version TensorFlow

conda install tensorflow-gpu=1.14

tips：

• “=”： Add “=” On behalf of automatically installing the latest updated version of the large version of the software package , Such as “tensorflow=1.13” It will be installed automatically tensorflow edition 1.13 The latest version of , namely “1.13.” Decimal point followed by the latest number

• “==”： Add after the package “==” On behalf of installing the specified version of the package , The version number after the double equal sign must be written completely , Such as “tensorflow==1.13.1” Is to install 1.13.1 Version of tensorflow

notes ： about 1.15 And earlier ,CPU and GPU Packages are separate ：
CPU edition

conda install tensorflow==1.14

4. Test the installation for success

python

import tensorflow as tf

tf._version_

Installation successful ！

verification gpu Is it available

tf.test.is_gpu_available()

Uninstall environment command

conda uninstall -n tf --all

yes

see conda Environmental Science

conda info -e

Reference tutorial ：

1. Installation tutorial ：https://zhuanlan.zhihu.com/p/84568790
2. tensorflow Official website ：https://tensorflow.google.cn/install/gpu