Intel® AI Containers

This repository contains Dockerfiles, scripts, yaml files, Helm charts, etc. used to scale out AI containers with versions of TensorFlow and PyTorch that have been optimized for Intel platforms. Scaling is done with python, Docker, kubernetes, kubeflow, cnvrg.io, Helm, and other container orchestration frameworks for use in the cloud and on-premise.

Project Setup

Define your project's registry each time you use the project:

export REGISTRY=<registry_name>

Set Up Docker Engine

You'll need to install Docker Engine on your development system. Note that while Docker Engine is free to use, Docker Desktop may require you to purchase a license. See the Docker Engine Server installation instructions for details.

Set Up Docker Compose

Ensure you have Docker Compose installed on your machine. If you don't have this tool installed, consult the official Docker Compose installation documentation.

DOCKER_CONFIG=${DOCKER_CONFIG:-$HOME/.docker}
mkdir -p $DOCKER_CONFIG/cli-plugins
curl -SL https://github.com/docker/compose/releases/download/v2.19.0/docker-compose-linux-x86_64 -o $DOCKER_CONFIG/cli-plugins/docker-compose
chmod +x $DOCKER_CONFIG/cli-plugins/docker-compose
docker compose version

Development Container

Alternatively, build and utilize the MLOps Development container rather than setting up docker compose. Docker Engine is still required and needs to be set up properly on the host system.

docker build -t intel/mlops:compose-devel \
  -f .github/utils/Dockerfile.compose \
  --pull .
cd <framework>
docker run --rm intel/mlops:compose-devel docker compose up --build

Build Ingredient Containers

Select your framework and run the docker compose commands:

cd <framework>
docker compose up --build

To configure ingredient containers differently, see the framework README for a table of options for each ingredient.

Recipe Integration

To include a new recipe start by creating a new stage in a given framework's Dockerfile.

FROM base_target AS my_recipe_target

RUN pip install -r requirements.txt
RUN ...

Create as many stages you want, but make sure to note your final target name. Then add a new service in the framework's docker-compose.yaml file.

service_name:
  image: ${REGISTRY}/aiops/mlops-ci:b-${GITHUB_RUN_NUMBER:-0}-${BASE_IMAGE_NAME:-ubuntu}-${BASE_IMAGE_TAG:-22.04}-${PACKAGE_OPTION:-pip}-py${PYTHON_VERSION:-3.10}-ipex-${IPEX_VERSION:-1.12.1}-my-package-${MY_PACKAGE_VERSION:-<version>}
  build:
    args:
      MY_PACKAGE_VERSION: ${MY_PACKAGE_VERSION:-<version>}
    target: my_recipe_target
  command: >
    sh -c "python -c 'import my_package; print(\"My Package Version:\", my_package.__version__)'"
  extends:
    service: base_target

For more information on how to customize your recipe, see the docker compose documentation.

Partial Layer Integration

When re-using the AI Ingredient Containers for Intel Architectures it is often more efficient to only use portions of the image rather than all of the layers. The provided examples copy only the python environment and command-line tools installed in the image.

For Stock Python:

COPY --from=intel/intel-optimized-tensorflow:<my_pip_tag> /usr/local/lib/python${PYTHON_VERSION}/dist-packages /usr/local/lib/python${PYTHON_VERSION}/dist-packages
COPY --from=intel/intel-optimized-tensorflow:<my_pip_tag> /usr/local/bin /usr/local/bin

For Intel Distribution for Python:

COPY --from=intel/intel-optimized-tensorflow:<my_idp_tag> /root/conda/envs/idp /root/conda/envs/<my_env>

Alter Existing Recipe

When changing a recipe, alter the framework's Dockerfile and then utilize GitHub Actions to build and test a framework. When changing a version, alter the framework's docker-compose.yaml file by modifying all instances of the version:

${INC_VERSION:-2.1.0} -> ${INC_VERSION:-2.1.1}

And then build and test using GitHub Actions.

Notebooks and MLFlow

Jupyter

When using a container intended to launch a jupyter notebook server, start the Jupyter Server via the docker run command and copy the URL (something like http://127.0.0.1:$PORT/?token=***) into your browser, the port is 8888 by default.

cd <framework>
docker compose build jupyter
docker compose run -d --rm <image-name> jupyter notebook --notebook-dir=/jupyter --ip 0.0.0.0 --no-browser --allow-root

MLFlow

Add an MLFLow Example:

• TensorFlow
• PyTorch
• SKLearn

Start the MLFlow server as a detached container and then re-use the container by executing a command in it.

export PORT=<myport>
cd <framework>
docker compose build mlflow
docker compose run -d --rm mlflow mlflow server -p $PORT -h 0.0.0.0
docker compose exec mlflow python /mlflow/myscript.py

Note: If you need to install more python packages to run any of the examples add a requirements.txt file to your working directory and append pip install -r /mlflow/requirements.txt into the docker compose exec command.

Access the results at https://localhost:<port>, the default port is 5000.

Troubleshooting

• See the Docker Troubleshooting Article.
• Verify that Docker Engine Post-Install Steps are completed.
• When facing socket error check the group membership of the user and ensure they are part of the docker group.
• After changing any docker files or configs, restart the docker service sudo systemctl restart docker.
• Enable Docker Desktop for WSL 2.
• If you are trying to access a container UI from the browser, make sure you have port forwarded and reconnect.
• If your environment requires a proxy to access the internet, export your development system's proxy settings to the docker environment:

export DOCKER_BUILD_ARGS="--build-arg ftp_proxy=${ftp_proxy} \
  --build-arg FTP_PROXY=${FTP_PROXY} --build-arg http_proxy=${http_proxy} \
  --build-arg HTTP_PROXY=${HTTP_PROXY} --build-arg https_proxy=${https_proxy} \
  --build-arg HTTPS_PROXY=${HTTPS_PROXY} --build-arg no_proxy=${no_proxy} \
  --build-arg NO_PROXY=${NO_PROXY} --build-arg socks_proxy=${socks_proxy} \
  --build-arg SOCKS_PROXY=${SOCKS_PROXY}"

export DOCKER_RUN_ENVS="-e ftp_proxy=${ftp_proxy} \
  -e FTP_PROXY=${FTP_PROXY} -e http_proxy=${http_proxy} \
  -e HTTP_PROXY=${HTTP_PROXY} -e https_proxy=${https_proxy} \
  -e HTTPS_PROXY=${HTTPS_PROXY} -e no_proxy=${no_proxy} \
  -e NO_PROXY=${NO_PROXY} -e socks_proxy=${socks_proxy} \
  -e SOCKS_PROXY=${SOCKS_PROXY}"

docker build $DOCKER_BUILD_ARGS -t my:tag .
docker run $DOCKER_RUN_ENVS --rm -it my:tag

Support

The Intel AI MLOps team tracks bugs and enhancement requests using GitHub issues. Before submitting a suggestion or bug report, search the existing GitHub issues to see if your issue has already been reported.

---

• Trademarks