Docker image to run Pegasus with TensorFlow 1.15 and CUDA 10

docker/Dockerfile

@@ -0,0 +1,57 @@
+FROM tensorflow/tensorflow:1.15.2-gpu
+
+ARG USER=pegasus
+ARG USER_UID=1000
+
+RUN apt-get update && apt-get install -y \
+	wget \
+	git \
+	man \
+	nano
+
+ENV CONDA_PATH /opt/conda
+ENV PATH ${CONDA_PATH}/bin:${PATH}
+
+RUN wget --quiet --no-check-certificate https://repo.anaconda.com/miniconda/Miniconda3-py37_4.8.3-Linux-x86_64.sh \
+    && echo "bb2e3cedd2e78a8bb6872ab3ab5b1266a90f8c7004a22d8dc2ea5effeb6a439a *Miniconda3-py37_4.8.3-Linux-x86_64.sh" | sha256sum -c - \
+    && /bin/bash Miniconda3-py37_4.8.3-Linux-x86_64.sh -f -b -p ${CONDA_PATH} \
+    && rm Miniconda3-py37_4.8.3-Linux-x86_64.sh \
+    && echo export PATH=${CONDA_PATH}/bin:'${PATH}' > /etc/profile.d/conda.sh
+
+RUN mkdir /src \
+    && git clone https://github.com/google-research/pegasus /src/
+
+RUN groupadd ${USER} && useradd -m -u ${USER_UID} -g ${USER} -s /bin/bash ${USER} \
+    && chown -R ${USER_UID} /src \
+    && chown -R ${USER_UID} ${CONDA_PATH}
+
+USER ${USER}
+
+# ENV PATH="/home/${USER}/.local/bin:$PATH"
+
+ARG python_version=3.6
+
+RUN conda config --append channels conda-forge
+RUN conda install -y \
+	python=${python_version} \
+	notebook \
+    && pip install --upgrade pip \
+    && pip install \
+	six==1.14.0 \
+	protobuf==3.12.0 \
+	cloudpickle==1.2.0 \
+	gsutil \
+    && pip install -r /src/requirements.txt
+
+RUN mkdir /src/ckpt/
+
+COPY docker_entrypoint.sh /src/
+
+COPY bash.bashrc /etc/
+
+WORKDIR /src
+
+ENV PYTHONPATH="/src/:$PYTHONPATH"
+
+ENTRYPOINT ["sh", "/src/docker_entrypoint.sh"]
+CMD ["sh"]

docker/Makefile

@@ -0,0 +1,23 @@
+help:
+	@cat Makefile
+
+IMAGE=google/pegasus
+PROJECT=pegasus
+DOCKER_FILE=Dockerfile
+SRC?=$(shell dirname `pwd`)
+DATA?="${HOME}/Data"
+
+build:
+	docker build -t $(IMAGE) -f $(DOCKER_FILE) .
+
+bash: build
+	docker run --init -it --gpus all --hostname $(PROJECT)-run -v $(SRC):/src/workspace -v $(DATA):/data --mount type=volume,source=$(PROJECT)-data,target=/src/ckpt/ --name $(PROJECT)-run $(IMAGE) bash
+
+ipython: build
+	docker run --init -it --gpus all --hostname $(PROJECT)-run -v $(SRC):/src/workspace -v $(DATA):/data --mount type=volume,source=$(PROJECT)-data,target=/src/ckpt/ --name $(PROJECT)-run $(IMAGE) ipython
+
+notebook: build
+	docker run --init -it --gpus all --hostname $(PROJECT)-run -v $(SRC):/src/workspace -v $(DATA):/data --mount type=volume,source=$(PROJECT)-data,target=/src/ckpt/ --net=host --name $(PROJECT)-run $(IMAGE) jupyter notebook --port=8888 --ip=0.0.0.0
+
+test: build
+	docker run --init -it --gpus all --hostname $(PROJECT)-run -v $(SRC):/src/workspace -v $(DATA):/data --mount type=volume,source=$(PROJECT)-data,target=/src/ckpt/ --name $(PROJECT) $(IMAGE) python3 pegasus/bin/train.py --params=aeslc_transformer --param_overrides=vocab_filename=ckpt/pegasus_ckpt/c4.unigram.newline.10pct.96000.model --train_init_checkpoint=ckpt/pegasus_ckpt/model.ckpt-1500000 --model_dir=ckpt/pegasus_ckpt/aeslc

docker/README.md

@@ -0,0 +1,41 @@
+# Using PEGASUS via Docker
+
+This directory contains the `Dockerfile` to make it easy to get up and running with
+PEGASUS via [Docker](http://www.docker.com/).
+
+## Installing Docker
+
+General installation instructions are
+[on the Docker site](https://docs.docker.com/installation/).
+
+## Installing GPU support
+
+For GPU support install NVIDIA drivers (ideally the latest) and the [NVIDIA Container Toolkit](https://github.com/NVIDIA/nvidia-docker).
+
+## Running the container
+
+`Makefile` is used to simplify docker commands within `make` commands.
+
+Build the container and start a bash
+
+    $ make bash
+
+Build the container and start an iPython shell
+
+	$ make ipython
+
+Build the container and start a Jupyter Notebook
+
+	$ make notebook
+
+Build the container and start the finetuning test described in the repository's README:
+
+	$ make test
+
+Mount a volume for external data sets
+
+    $ make DATA=~/mydata
+
+Prints all make tasks
+
+    $ make help

docker/bash.bashrc

@@ -0,0 +1,39 @@
+#
+# /etc/bash.bashrc
+#
+
+export PS1="\[\e[31m\]pegasus-docker\[\e[m\] \[\e[33m\]\w\[\e[m\] > "
+export TERM=xterm-256color
+alias grep="grep --color=auto"
+alias ls="ls --color=auto"
+
+echo -e "\e[1;31m"
+cat<<P
+__________                                         
+\______   \ ____   _________    ________ __  ______
+ |     ___// __ \ / ___\__  \  /  ___/  |  \/  ___/
+ |    |   \  ___// /_/  > __ \_\___ \|  |  /\___ \ 
+ |____|    \___  >___  (____  /____  >____//____  >
+               \/_____/     \/     \/           \/ 
+
+P
+echo -e "\e[0;33m"
+
+if [[ $EUID -eq 0 ]]; then
+  cat <<WARN
+WARNING: You are running this container as root, which can cause new files in
+mounted volumes to be created as the root user on your host machine.
+
+To avoid this, run the container by specifying your user's userid:
+
+$ docker run -u \$(id -u):\$(id -g) args...
+WARN
+else
+  cat <<EXPL
+You are running this container as user with ID $(id -u) and group $(id -g),
+which should map to the ID and group for your user on the Docker host. Great!
+EXPL
+fi
+
+# Turn off colors
+echo -e "\e[m"

docker/docker_entrypoint.sh

@@ -0,0 +1,8 @@
+#!/bin/sh
+
+if [ ! -d /src/ckpt/pegasus_ckpt/ ]
+then
+    gsutil cp -r gs://pegasus_ckpt /src/ckpt/
+fi
+
+exec "$@"