tensorflow-in-windows (1/2)

tensorflow-in-windows (1/2)

• this post covers
  • install python using anaconda
  • install tensorflow-cpu, which supports cpu only
  • install vs-code, which can be used for editor
• next post covers
  • install cuda
  • install tensorflow-gpu, which supports gpu too
• environment
  • Windows / 10

---

anaconda-and-python

• it is used for encapsulating python environment
  • like the purpose we are using vm-ware
• when you install python directly on the system, it is hard to add other version of python
  • the PATH environment variable can be conflicted
• in this situation, we gotta create virtual environment for each version of python
  • the environments do not affect each other

• visit here to download the installer for windows
  • do not forget to click "Windows" button
  • version of python does not matter when you choose the installer
  • if your system is Windows 10 64 bit, download the "64-Bit Graphical Installer" of any version of python

• execute the installer and follow default options

• when the installation is over, click the search bar
  • type "anaconda prompt" and run it

• you can see the prompt with "(base)"
  • anaconda prompt displays the environment name which you are using
  • the version of python would be that in the installer description

• now let us create new environment with python 3.5
  • type "conda create -n Python35 python=3.5"
  • this will make an environment with its name is "Python35"

• when the task done, you can checkout the environment
  • type "activate Python35"
  • now you can see the prompt with "(Python35)"
  • the version of python would be 3.5.X

• if you return to previous environment, type "deactivate"
  • the "activate" and "deactivate" works like stack

tensorflow

• if you do not have visual-studio-ide, visit here and download the appropriate redist.exe
  • if your system is Windows 10 64 bit, download the "x64: vc_redist.x64.exe"
  • execute the redist.exe

• opened the anaconda prompt, enter the environment Python35
  • type "pip install tensorflow"
  • this will install the latest tensorflow-cpu fits in current python

• run the python and check the version of tensorflow
  • you can check whether the tensorflow supports gpu with "is_gpu_available()"
  • now you successfully installed tensorflow-cpu

vs-code

• though there are plenty of python editors, I recommend you the vs-code
  • it supports anaconda well
  • it runs python with light weight

• visit here to download the installer for windows
  • click the download button

• execute the installer and follow default options

• when the installation is over, click the search bar
  • find the "Visual Studio Code" and run it

• click the extension tab
  • type "python" and find the python extension
  • click the install button

• when it installed, open any workspace
  • make an empty python file

 

• now you can see current environment at the left-bottom
  • whenever you compile the python file, the python will be used

• check the import on tensorflow
  • if you did not install the tensorflow at base, debugging would fail

• click the left-bottom and change the environment
  • we installed tensorflow at Python35
  • you can see the debugging successfully done