dockerfile.txt

ARG BASE_IMAGE=nvcr.io/nvidia/pytorch:19.07-py3
FROM $BASE_IMAGE
ARG BASE_IMAGE
RUN echo "Installing Apex on top of ${BASE_IMAGE}"

RUN apt-get update && apt-get install wget -y
RUN mkdir ~/Downloads
RUN wget -P ~/Downloads https://repo.continuum.io/archive/Anaconda3-2019.10-Linux-x86_64.sh

RUN sh ~/Downloads/Anaconda3-2019.10-Linux-x86_64.sh -b -p /home/anaconda3
ENV PATH="/home/anaconda3/bin:$PATH"
RUN echo $PATH
RUN conda install pytorch=1.4.0 torchvision cudatoolkit=10.1 -c pytorch

RUN apt-get install gcc g++ vim -y
RUN pip install tensorboardX
ENV PYTHONIOENCODING="utf8"
RUN conda install nltk
RUN python -m nltk.downloader punkt
ENV DEBIAN_FRONTEND=noninteractive
RUN apt-get install -y less tzdata
ENV TZ America/Los_Angeles

RUN pip install regex
RUN pip install virtualenv
RUN pip install boto3
RUN pip install scikit-learn
RUN pip install comet_ml
RUN pip install transformers==2.11.0

# configure apex
# make sure we don't overwrite some existing directory called "apex"
WORKDIR /tmp/unique_for_apex
# uninstall Apex if present, twice to make absolutely sure :)
RUN pip uninstall -y apex || :
RUN pip uninstall -y apex || :
# SHA is something the user can touch to force recreation of this Docker layer,
# and therefore force cloning of the latest version of Apex
RUN SHA=ToUcHMe git clone https://github.com/NVIDIA/apex.git
WORKDIR /tmp/unique_for_apex/apex
RUN pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" .
WORKDIR /workspace

# configure fairseq
WORKDIR /tmp/unique_for_fairseq
# SHA is something the user can touch to force recreation of this Docker layer,
# and therefore force cloning of the latest version of Apex
RUN SHA=ToUcHMe git clone https://github.com/pytorch/fairseq
WORKDIR /tmp/unique_for_fairseq/fairseq
RUN pip install --editable ./

WORKDIR /workspace