Dockerfile

# prepare system for Ubuntu 16.04 with python3.6
FROM ubuntu:16.04
FROM nvidia/cuda:10.0-cudnn7-runtime-ubuntu16.04
# Set user as root so next commands run correctly
USER root

# install python3.6 and python3.6-dev
RUN apt-get update
RUN apt-get install -y software-properties-common
RUN add-apt-repository ppa:deadsnakes/ppa
RUN apt-get update
RUN apt-get install -y apt-utils build-essential python3.6 python3.6-dev python3-pip python3.6-venv cmake make wget git

# update pip
RUN python3.6 -m pip install pip --upgrade
RUN python3.6 -m pip install wheel
RUN pip3 install --upgrade setuptools

# install pytorch
RUN pip3 install torch==1.1.0 torchvision==0.3.0 -f https://download.pytorch.org/whl/cu100/torch_stable.html

# install tensorflow 1.13.1
RUN pip3 install tensorflow-gpu==1.13.1

RUN mkdir -p /app/

RUN git clone https://github.com/facebookresearch/ParlAI.git /app/ParlAI
RUN cd /app/ParlAI; pip3 install -r ./requirements.txt; echo "" > README.md; python3.6 setup.py develop

RUN mkdir -p /app/hrinlp
COPY ./requirements.txt /app/hrinlp/

RUN pip3 install -r /app/hrinlp/requirements.txt

RUN ln -s /usr/local/bin/python3 /usr/bin/python & \
    ln -s /usr/local/bin/pip3 /usr/bin/pip

RUN apt-get install -y nano

COPY ./README.md /app/hrinlp/
COPY ./data/barista /app/hrinlp/data/barista
COPY ./data/barista-personalised /app/hrinlp/data/barista-personalised
COPY ./data/barista-personalised-order /app/hrinlp/data/barista-personalised-order
COPY ./data/ /app/ParlAI/data

COPY ./baselines /app/hrinlp/baselines
COPY ./parlai_internal/ /app/ParlAI/parlai_internal
COPY ./parlai_internal/scripts/params.py /app/ParlAI/parlai/core/params.py