Jupyter_Chainer Dockerfile

Dockerfile

FROM jupyter/minimal

# referenced : https://github.com/jupyter/docker-demo-images
USER jovyan
ENV HOME /home/jovyan
ENV SHELL /bin/bash
ENV USER jovyan
ENV PATH $CONDA_DIR/bin:$CONDA_DIR/envs/python2/bin:$PATH
WORKDIR $HOME

USER jovyan

RUN conda install --yes numpy pandas scikit-learn scikit-image matplotlib scipy seaborn sympy cython patsy statsmodels cloudpickle dill numba bokeh && conda clean -yt

RUN conda create -p $CONDA_DIR/envs/python2 python=2.7 ipython numpy pandas scikit-learn scikit-image matplotlib scipy seaborn sympy cython patsy statsmodels cloudpickle dill numba bokeh && conda clean -yt
RUN $CONDA_DIR/envs/python2/bin/python $CONDA_DIR/envs/python2/bin/ipython kernelspec install-self --user

# Install CUDA
USER root
RUN apt-get purge nvidia*
ENV CUDA_RUN http://developer.download.nvidia.com/compute/cuda/7_0/Prod/local_installers/cuda_7.0.28_linux.run
RUN apt-get update && apt-get install -q -y \
  wget \
  build-essential
RUN cd /opt && \
  wget $CUDA_RUN && \
  chmod +x *.run
RUN cd /opt && mkdir nvidia_installers && \
  ./cuda_7.0.*.run -extract=`pwd`/nvidia_installers
RUN apt-get install -y module-init-tools
RUN cd /opt/nvidia_installers && \
  ./NVIDIA-Linux-x86_64-*.run -s -N --no-kernel-module

RUN cd /opt/nvidia_installers && \
  ./cuda-linux64-rel-7.0*.run -noprompt

ENV CUDA_HOME=/usr/local/cuda-7.0
ENV LD_LIBRARY_PATH=${CUDA_HOME}/lib64

ENV PATH=${CUDA_HOME}/bin:${PATH}

# Install Chainer
USER jovyan
RUN /opt/conda/envs/python2/bin/pip install chainer chainer-cuda-deps
RUN git clone https://github.com/pfnet/chainer.git

CMD ipython notebook

run

$ docker build -t jupyter_chainer .
$ docker run -d -it \
  --device /dev/nvidia0:/dev/nvidia0 \
  --device /dev/nvidia1:/dev/nvidia1 \ 
  --device /dev/nvidia2:/dev/nvidia2 \
  --device /dev/nvidiactl:/dev/nvidiactl \
  --device /dev/nvidia-uvm:/dev/nvidia-uvm \
  -p 8888:8888 \
  -v /path/to/notbookedata/:/home/jovyan/notebookdata \
  jupyter_chainer