I created a simple machine learning Docker container image for executing small experiments on AWS and Azure.
The container uses Ubuntu 18.04.2 LTS at its core with Python 3 and common Python ML libraries:
- pandas
- scipy
- numpy
- matplotlib
- records
- tensorflow
Machine Learning with Docker Source Code
#Download base image Ubuntu 18.04.2 LTS
FROM ubuntu:latest
# Update Ubuntu Software repository
RUN apt-get update && apt-get -y upgrade
RUN apt-get install -y sudo curl lsb-release wget
# Install Python 3 with pip
RUN apt-get install -y python3-pip
COPY requirements.txt .
RUN pip3 install -r requirements.txt
# Environment settings
ENV PYTHONUNBUFFERED 1
RUN echo 'alias py=python3' >> ~/.bashrc
RUN echo 'alias pip=pip3' >> ~/.bashrc
# TODO: Add nvidia-docker once it supports Windows
# https://github.com/NVIDIA/nvidia-docker/issues/429