Tensorflow using virtual environments

Tensorflow 2.6(Current version 2.11 at 2023-03-23)

$ sudo apt-get install python3-venv

$ python3 -m venv envtf26

$ source envtf26/bin/activate

(envtf26) cloud@AsusDesktop:~/Gaia$
 

(envtf26) cloud@AsusDesktop:~/Gaia$ pip list  
Package       Version
------------- -------
pip           20.0.2  
pkg-resources 0.0.0   
setuptools    44.0.0

(envtf26) cloud@AsusDesktop:~/Gaia$ pip install tensorflow==2.6

(envtf26) cloud@AsusDesktop:~/Gaia$ pip install --upgrade protobuf==3.20

(envtf26) cloud@AsusDesktop:~/Gaia$ pip install keras==2.6

(envtf26) cloud@AsusDesktop:~/Gaia$ python3           
Python 3.8.10 (default, Mar 13 2023, 10:26:41)  
[GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf      
>>> print(tf.keras.__version__)  
2.6.0

(envtf26) cloud@AsusDesktop:~/Gaia$ pip install matplotlib==3.6 #if you install without 3.6 will install 3.7 and it will break tensorflow installation

References:

[1] matplotlib compa https://matplotlib.org/stable/devel/min_dep_policy.html