mirror of
https://github.com/bspeice/neuralnets-docker
synced 2024-12-04 12:48:09 -05:00
42 lines
1.3 KiB
Docker
42 lines
1.3 KiB
Docker
FROM nvidia/cuda:8.0-cudnn5-devel-ubuntu16.04
|
|
MAINTAINER Bradlee Speice
|
|
|
|
LABEL Description="Jupyter server setup for ECBM E4040 Neural Networks" Version="0.3"
|
|
|
|
# Update our cache first
|
|
RUN apt-get update && \
|
|
|
|
# Install packages needed for a sane linux system
|
|
apt-get install -y git && \
|
|
|
|
# Don't install broken Pip
|
|
apt-get install -y python-pip=8.1.1-2 python3-pip=8.1.1-2 python-pip-whl=8.1.1-2 && \
|
|
|
|
# Install the Scipy stuff we need
|
|
apt-get install -y \
|
|
python3 libpython3-dev \
|
|
python-pandas python-matplotlib python-sklearn python-pyodbc \
|
|
python3-pandas python3-matplotlib python3-sklearn python3-pyodbc \
|
|
texlive-latex-extra texlive-fonts-recommended texlive-generic-recommended pandoc
|
|
|
|
# And the python-specific tools
|
|
RUN pip install theano jupyter librosa mir_eval && \
|
|
pip3 install theano jupyter librosa mir_eval && \
|
|
ipython2 kernel install
|
|
|
|
# And the startup script
|
|
COPY . /
|
|
|
|
# Set up Theano for the GPU
|
|
ENV THEANO_FLAGS='floatX=float32,device=gpu'
|
|
|
|
# Set up an unprivileged user to run as
|
|
RUN useradd jupyter -s /bin/false && \
|
|
mkdir /home/jupyter && \
|
|
chown jupyter:jupyter /home/jupyter && \
|
|
passwd jupyter -l
|
|
|
|
ENTRYPOINT ["/sbin/runuser", "-u", "jupyter", "/usr/local/bin/start_jupyter"]
|
|
|
|
EXPOSE 8888
|