deep-docker

This repository contains various Dockerfiles for my deep learning research. The built docker images can be started using nvida-docker.

The base image for all Dockerfiles is nvidia/cuda. The docker hub site can be found here.

Building images and starting containers

To build any of the provided Dockerfiles go the relevant folder and execute:

docker build -t <desired-name-of-image> .

The image can be executed via nvidia-docker:

nvidia-docker run -it --rm <desired-name-of-image>

Index

Currently, the following Dockerfiles exist:

Keras

The Dockerfile for Keras uses CUDA in version 10.0 and cudnn in version 7. Thus, TensorFlow in version 1.8 can be used in combination with Keras.

Other useful libraries are also installed:

keras-contribu
scikit-learn
h5py
gensim
pandas

`pipenv`

The Dockerfile for pipenv and pyenv uses CUDA in version 10.0.

Here's a small pyenv example that installs Python 3.7:

root@5bc18bc97c62:/# pyenv install 3.7.0
Downloading Python-3.7.0.tar.xz...
-> https://www.python.org/ftp/python/3.7.0/Python-3.7.0.tar.xz
Installing Python-3.7.0...
Installed Python-3.7.0 to /.pyenv/versions/3.7.0
root@5bc18bc97c62:/# pyenv global 3.7.0
root@5bc18bc97c62:/# python --version
Python 3.7.0

`pytorch`

The Dockerfile for pytorch 1.1.0 uses CUDA in version 10.1.

stefan-it / deep-docker