Deepo is a series of Docker images that

• allows you to quickly set up your deep learning research environment
• supports almost all commonly used deep learning frameworks
• supports GPU acceleration (CUDA and cuDNN included), also works in CPU-only mode
• works on Linux (CPU version/GPU version), Windows (CPU version) and OS X (CPU version)

and their Dockerfile generator that

• allows you to customize your own environment with Lego-like modules
• automatically resolves the dependencies for you

---

Table of contents

• Quick Start
  • GPU Version
    • Installation
    • Usage
  • CPU Version
    • Installation
    • Usage
• Customization
  • Unhappy with all-in-one solution?
  • Other python versions
  • Jupyter support
  • Build your own customized image
• Comparison to Alternatives
• Available Tags
• Contributing
• Licensing

---

Quick Start

GPU Version

Installation

Step 1. Install Docker and nvidia-docker.

Step 2. Obtain the all-in-one image from Docker Hub

docker pull ufoym/deepo

Usage

Now you can try this command:

nvidia-docker run --rm ufoym/deepo nvidia-smi

This should work and enables Deepo to use the GPU from inside a docker container. If this does not work, search the issues section on the nvidia-docker GitHub -- many solutions are already documented. To get an interactive shell to a container that will not be automatically deleted after you exit do

nvidia-docker run -it ufoym/deepo bash

If you want to share your data and configurations between the host (your machine or VM) and the container in which you are using Deepo, use the -v option, e.g.

nvidia-docker run -it -v /host/data:/data -v /host/config:/config ufoym/deepo bash

This will make /host/data from the host visible as /data in the container, and /host/config as /config. Such isolation reduces the chances of your containerized experiments overwriting or using wrong data.

Please note that some frameworks (e.g. PyTorch) use shared memory to share data between processes, so if multiprocessing is used the default shared memory segment size that container runs with is not enough, and you should increase shared memory size either with --ipc=host or --shm-size command line options to nvidia-docker run.

nvidia-docker run -it --ipc=host ufoym/deepo bash

CPU Version

Installation

Step 1. Install Docker.

Step 2. Obtain the all-in-one image from Docker Hub

docker pull ufoym/deepo:cpu

Usage

Now you can try this command:

docker run -it ufoym/deepo:cpu bash

If you want to share your data and configurations between the host (your machine or VM) and the container in which you are using Deepo, use the -v option, e.g.

docker run -it -v /host/data:/data -v /host/config:/config ufoym/deepo:cpu bash

This will make /host/data from the host visible as /data in the container, and /host/config as /config. Such isolation reduces the chances of your containerized experiments overwriting or using wrong data.

Please note that some frameworks (e.g. PyTorch) use shared memory to share data between processes, so if multiprocessing is used the default shared memory segment size that container runs with is not enough, and you should increase shared memory size either with --ipc=host or --shm-size command line options to docker run.

docker run -it --ipc=host ufoym/deepo:cpu bash

You are now ready to begin your journey.

$ python

>>> import tensorflow
>>> import sonnet
>>> import torch
>>> import keras
>>> import mxnet
>>> import cntk
>>> import chainer
>>> import theano
>>> import lasagne
>>> import caffe
>>> import caffe2

$ caffe --version

caffe version 1.0.0

$ th

 │  ______             __   |  Torch7
 │ /_  __/__  ________/ /   |  Scientific computing for Lua.
 │  / / / _ \/ __/ __/ _ \  |  Type ? for help
 │ /_/  \___/_/  \__/_//_/  |  https://github.com/torch
 │                          |  http://torch.ch
 │
 │th>

Customization

Note that docker pull ufoym/deepo mentioned in Quick Start will give you a standard image containing all available deep learning frameworks. You can customize your own environment as well.

Unhappy with all-in-one solution?

If you prefer a specific framework rather than an all-in-one image, just append a tag with the name of the framework. Take tensorflow for example:

docker pull ufoym/deepo:tensorflow

Other python versions

Note that all python-related images use Python 3.6 by default. If you are unhappy with Python 3.6, you can also specify other python versions:

docker pull ufoym/deepo:py27

docker pull ufoym/deepo:tensorflow-py27

Currently, we support Python 2.7 and Python 3.6.

See Available Tags for a complete list of all available tags. These pre-built images are all built from docker/Dockerfile.* and circle.yml. See How to generate docker/Dockerfile.* and circle.yml if you are interested in how these files are generated.

Jupyter support

Step 1. pull the image with jupyter support

docker pull ufoym/deepo:all-py36-jupyter

Note that the tag could be either of all-py36-jupyter, py36-jupyter, all-py27-jupyter, or py27-jupyter.

Step 2. run the image

nvidia-docker run -it -p 8888:8888 ufoym/deepo:all-py36-jupyter jupyter notebook --no-browser --ip=0.0.0.0 --allow-root --NotebookApp.token= --notebook-dir='/root'

Build your own customized image with Lego-like modules

Step 1. prepare generator

git clone https://github.com/ufoym/deepo.git
cd deepo/generator
pip install -r requirements.txt

Step 2. generate your customized Dockerfile

For example, if you like pytorch and lasagne, then

python generate.py Dockerfile pytorch lasagne

This should generate a Dockerfile that contains everything for building pytorch and lasagne. Note that the generator can handle automatic dependency processing and topologically sort the lists. So you don't need to worry about missing dependencies and the list order.

You can also specify the version of Python:

python generate.py Dockerfile pytorch lasagne python==3.6

Step 3. build your Dockerfile

docker build -t my/deepo .

This may take several minutes as it compiles a few libraries from scratch.

Comparison to alternatives

.	modern-deep-learning	dl-docker	jupyter-deeplearning	Deepo
ubuntu	16.04	14.04	14.04	16.04
cuda	❌	8.0	6.5-8.0	9.0
cudnn	❌	v5	v2-5	v7
theano	❌	✔️	✔️	✔️
tensorflow	✔️	✔️	✔️	✔️
sonnet	❌	❌	❌	✔️
pytorch	❌	❌	❌	✔️
keras	✔️	✔️	✔️	✔️
lasagne	❌	✔️	✔️	✔️
mxnet	❌	❌	❌	✔️
cntk	❌	❌	❌	✔️
chainer	❌	❌	❌	✔️
caffe	✔️	✔️	✔️	✔️
caffe2	❌	❌	❌	✔️
torch	❌	✔️	✔️	✔️

---

Available Tags

.	GPU / Python 3.6	GPU / Python 2.7	CPU-only / Python 3.6	CPU-only / Python 2.7
all-in-one	all-py36 all py36 latest	all-py27 py27	all-py36-cpu all-cpu py36-cpu cpu	all-py27-cpu py27-cpu
all-in-one with jupyter	all-py36-jupyter py36-jupyter	all-py27-jupyter py27-jupyter	all-py36-jupyter-cpu py36-jupyter-cpu	all-py27-jupyter-cpu py27-jupyter-cpu
theano	theano-py36 theano	theano-py27	theano-py36-cpu theano-cpu	theano-py27-cpu
tensorflow	tensorflow-py36 tensorflow	tensorflow-py27	tensorflow-py36-cpu tensorflow-cpu	tensorflow-py27-cpu
sonnet	sonnet-py36 sonnet	sonnet-py27	sonnet-py36-cpu sonnet-cpu	sonnet-py27-cpu
pytorch	pytorch-py36 pytorch	pytorch-py27	pytorch-py36 pytorch	pytorch-py27
keras	keras-py36 keras	keras-py27	keras-py36-cpu keras-cpu	keras-py27-cpu
lasagne	lasagne-py36 lasagne	lasagne-py27	lasagne-py36-cpu lasagne-cpu	lasagne-py27-cpu
mxnet	mxnet-py36 mxnet	mxnet-py27	mxnet-py36-cpu mxnet-cpu	mxnet-py27-cpu
cntk	cntk-py36 cntk	cntk-py27	cntk-py36-cpu cntk-cpu	cntk-py27-cpu
chainer	chainer-py36 chainer	chainer-py27	chainer-py36-cpu chainer-cpu	chainer-py27-cpu
caffe	caffe-py36 caffe	caffe-py27	caffe-py36-cpu caffe-cpu	caffe-py27-cpu
caffe2	caffe2-py36 caffe2	caffe2-py27	caffe2-py36-cpu caffe2-cpu	caffe2-py27-cpu
torch	torch	torch	torch-cpu	torch-cpu

---

Contributing

We appreciate all contributions. If you are planning to contribute back bug-fixes, please do so without any further discussion. If you plan to contribute new features, utility functions or extensions, please first open an issue and discuss the feature with us.

Licensing

Deepo is MIT licensed.