Tensorflow actual combat Google deep learning framework version 2 learning summary tensorflow installation

The installation method is mainly divided into two versions , Namely CPU and GPU edition ,CPU The version is relatively simple , and GPU The version process is complicated . If you are just a novice tensorflow There is no need to install GPU edition , When you need to build dozens of layers of network, install GPU edition , meanwhile GPU The installation of version is different for each computer .
CPU Version main reference ：
https://blog.csdn.net/colourful_sky/article/details/73823939
https://blog.csdn.net/colourful_sky/article/details/73823939
https://blog.csdn.net/zzc15806/article/details/73662491
https://blog.csdn.net/WIinter_FDd/article/details/66523468
https://blog.csdn.net/u014680339/article/details/72459392
https://blog.csdn.net/hgdwdtt/article/details/78633232
Installation environment ：Ubuntu14.04/16.04LTS
1. install anaconda
1.1 Download the installation file from the official website of Tsinghua University https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/, Here I choose Anaconda3-5.1.0-Linux-x86_64.sh, Note that there Python The version is Python3.6, at present anaconda There are only two versions , Namely Python3.6 and Python2.7 edition , I suggest you to follow the tide of the times and choose a higher version Python3.6,
1.2 Run the download directly sh File can ：

bash Anaconda3-5.1.0-Linux-x86_64.sh

1.3 After the installation is completed, you must verify whether the installation is successful , start-up anaconda, If it starts normally, the installation is successful , There is usually no problem with this step , If it doesn't start , Reloading won't solve the problem , It is recommended to reinstall the system

anaconda-navigator

2. install tensorflow
2.1 establish tensorflow Environmental Science

conda create -n tensorflow python=3.6

Installed here Python yes 3.6 edition ,anaconda and Python as well as tensorflow We have to match it up
2.2 Activate tensorflow Environmental Science

source activate tensorflow

After successful activation , There will be a in front of the terminal （tensorflow） identification
2.3 install tensorflow

pip install –ignore-installed –upgrade https://mirrors.tuna.tsinghua.edu.cn/tensorflow/linux/cpu/tensorflow-1.4.0-cp36-cp36m-linux_x86_64.whl

What I installed here is tensorflow1.4.0 Of CPU edition , Corresponding to the edition of the book , If you want to install other versions, you can go to the website of Tsinghua University to find https://mirrors.tuna.tsinghua.edu.cn/help/tensorflow/
2.4 verification tensorflow Is the installation successful

source activate tensorflow # Activate first tensorflow Environmental Science
python # start-up python Environmental Science
import tensorflow as tf
hello = tf.constant(‘Hello, TensorFlow!’)
sess = tf.Session()
print (sess.run(hello))
Hello, TensorFlow! # Congratulations ！ Installation successful ！

3. stay anaconda Install the code editor on
anaconda Several code editors are required in , We choose jupyter, Interactive notebooks
3.1 First start anaconda

anaconda-navigator

start-up anaconda After that, it can be installed directly inside the software jupyter, After installation, you can directly click to open anaconda, You will see in Firefox jupyter Software
3.2 stay jupyter Use in tensorflow
First activate... In the terminal tensorflow Environmental Science

source activate tensorflow # Activate Tensorflow Environmental Science

Then open the anaconda Software

anaconda-navigator

Now open jupyter You can write code directly , If not activated tensorflow, stay jupyter Written in tensorflow The following error occurs in the code ：

import tensorflow as tf # Using this code, the following error will appear
modulenotfounderror no module named tensorflow