This repository consists of scripts and Dockerfile to setup an environment necessary to follow machine learning course from http://course.fast.ai. It is tested on AWS GPU instance (p2.xlarge) using Ubuntu 16.04 OS.
The only requirement is an Ubuntu 16.04 OS with GPU.
- Clone this repo and move to repo directory,
git clone https://github.com/GDP-ADMIN/fastai-course && cd fastai-course
. - Run ubuntu.sh (
./ubuntu.sh
) script to install NVidia cuda, docker-ce, and nvidia-docker2. - Close current terminal and open a new terminal. This is necessary so you can run docker command without sudo.
- Build docker image,
./fastai.sh build
- Create docker container to start fastai env,
PASSWORD=$PASSWORD ./fastai.sh up
. Change the $PASSWORD variable to something random since it is used as password when accessing jupyter notebook. Open http://[machine ip]:8888/ from your browser to visit jupyter notebook.
The ./fastai.sh up
command binds mount nbs directory from host to container.
I suggest to save your notebook file under nbs directory.
Before deleting the instance, i suggest to copy nbs directory to other location.