Tensorflow Installation Guide

This is a Tensorflow installation guide which will help to setup Tensorflow with GPU support in a windows machine(windows 10 and 11).

Table of contents

• Tensorflow Installation Guide
• Important
• Prerequiste
• Installation
• Error Guide
• Contact & support
• License

Important

Tensorflow for GPU support was removed after TF 2.10 version. The last supporting version of Tensorflow for GPU is 2.10 and below. Further documentation regarding the same is given in their official installation document. If you are not familiar with tensorflow, it is recommended to read this document before proceeding further

Prerequisite

You need python installed in your system. The following installation will be successfull with python version 3.10 to 3.7 . For GPU support , your need to download cuda 11.2 and its respective cudNN files.

• CUDA 11.2
• cudNN 8.1

The notebook contains the same documentation on the requirements. For the necessary dependecies required for the installation, you can go through the requirements.txt file. For smooth installation, do not use the requirements.txt for installing the dependencies before hand.

Installation

Clone this repository

git clone https://github.com/pratham-ak2004/Tensorflow-installation-guide

Navigate the the repository

cd Tensorflow-installation-guide

Using command prompt, create local python environment with version 3.10.11 . Replace python with the path of the interpreter with the required python version and env_name with your virtual environment name.

python -m venv env_name

Activate the python environment

.\env_name\scripts\activate ## for windows

Open the folder in jupyter notebook or any IDE

pip install jupyter-notebook ipykernel
python -m ipykernel install --user --name=env_name
jupyter notebook

Run the cells in Tensorflow installation.ipnb .

The installation is successfull when you get the below output when running the verification script

Error

There is a possibility that you may encounter a number of errors while running the notebook. It is recommended to refer Stackoverflow for solving these error.

Error: No module named ‘xxxxxx’
Solution: Install that module
Note the name of the module will not always equal the package name

!pip install xxxxx

For further errors you can contact the owner of this repository.

Drawbacks

Tensorflow 2.10 was released on September 06, 2022. Because of this it might not be compatible with some of the packages and you might encounter warnings. This is normal during installation. For more details refer the original documentation provided by Tensorflow

Contact & Support

Feel free to contact me : )

License