TensorFlow + GPU + Docker + Ubuntu 20.04 + Jupyter

In this tutorial we are going to discuss about running tensorlfow on GPU using docker containers and use jupyter notebooks as DL model development environment.

 

Benefits of using nvidia docker containers

1. Setting up CUDA, cudnn etc. is huge pain 

2. Docker keep your development environment isolated from host system and it is easily reproducible on any other machine

 

Once you have docker installed on your machine you need to install nvidia container toolkit

https://github.com/NVIDIA/nvidia-docker

After this run a tensorflow jupyter notebook with GPU support

https://hub.docker.com/r/tensorflow/tensorflow/