How to Setup Conda Environments and Add Kernels for Jupyter Notebook?

You want to experiment and create deep neural networks with PyTorch, TensorFlow 2.3, and TensorFlow 2.4. What will be an excellent way to create separate environments that will help better collaborate with other team members, manage dependencies, and isolate the different project environments?

The simplest way is to create and manage Conda environments and then add the Kernels to Jupyter notebook to access these environments.

Read here to learn more about Conda environments and the difference between Conda environments and Virtual environments.

What is a Conda environment, and why is it the most preferred method to manage different environments?

Conda lets you create, export, list, share, remove and update environments with different versions of Python and/or packages installed in them.

Conda environments are directories containing a specified collection of packages required for the isolated environment. It also allows you to select the exact version for the libraries you need for your environment.

Conda environments are language agnostic allowing you to manage Python as well as R packages.

The isolated environments created using Conda can be easily shared with other team members and used for your production move without worries about the version mismatch or missing dependencies.