Centos installation and configuration keras version

Centos installation and configuration keras version
Centos version:

1. Install theano

   1.1 download theano's zip file [https://github.com/theano/theano#, decompress it ~ /Site-packages/theano directory and name it theano
   1.2 command line input:

       python setup.py develop
2. Install Keras

   2.1 Download The keras zip file [https://github.com/fchollet/keras.git.pdf, decompress it ~ /Site-packages/keras directory, and name it keras 2.2 command line input:

       python setup.py install
3. Install the g ++ Environment

   On the command line, enter:

       yum install gcc-c++

   In centos, the g ++ installation package name is gcc-c ++.

4. Install other python packages
   sudo yum -y install numpy sudo yum -y install scipy
5. The default backend of keras is tensorflow, which can be converted to theano.

   5.1 open the. keras folder in the/root directory and enter the following command line:

       cd /root/.keras

   Pay special attention to the configuration in the root directory, instead of opening the keras. json file in keras 5.2, and entering in the command line:

       gedit keras.json

   Change "backend": "tensorflow" to "backend": "theano". You can set the backend to switch between theano and tensorflow. keras folder, no keras. json file, which can be created and the written content is:

       {    "image_dim_ordering": "th",     "epsilon": 1e-07,     "floatx": "float32",     "backend": "theano"}

Notes: Pythonsetup. py install and python setup. difference of py develop: python setup. py install is usually a third-party package python setup. py develop is a package that requires frequent code editing, but does not need to be re-installed. It is very convenient to edit the code to take effect immediately after it is installed in the (virtual) environment.