Pre-built docker images are available at https://hub.docker.com/r/dmlc/mxnet/

1. First pull the pre-built image

   docker pull dmlc/mxnet

2. Then we can run the python shell in the docker

   docker run -ti dmlc/mxnet python

   For example

   $ docker run -ti dmlc/mxnet python
   Python 2.7.6 (default, Jun 22 2015, 17:58:13)
   [GCC 4.8.2] on linux2
   Type "help", "copyright", "credits" or "license" for more information.
   >>> import mxnet as mx
   import mxnet as mx
   >>> quit()
   quit()

   Note: One may get the error message libdc1394 error: Failed to initialize libdc1394, which is due to opencv and can be ignored.

3. Train a model on MNIST to check everything works

   docker run dmlc/mxnet python /mxnet/example/image-classification/train_mnist.py
   

If the host machine has Nvidia CPUs, we can use dmlc/mxnet:gpu, which has both CUDA and CUDNN installed. To launch the docker, we need to install nvidia-docker first.

1. Pull the image

   docker pull dmlc/mxnet:cuda

2. Train MNIST on CPU 0

   nvidia-docker run dmlc/mxnet:cuda python /mxnet/example/image-classification/train_mnist.py --gpus 0

docker build -t dmlc/mxnet:cpu cpu
docker build -t dmlc/mxnet:cuda cuda