Notebook subdirectory - for use when launching a notebook instance from the SageMaker console

• lifecycle_configuration.sh

  Code of the lifecycle configuration script to launch a sagemaker Notebook instance

Studio subdirectory - for use when working within SageMaker Studio Classic

• example.ipynb

  Jupyter notebook with commands to demonstrate the following Artifactory and Sagemaker integrations after creating the notebook in an environment started with the lifecycle configuration script:

  1. JFrog CLI
  2. Artifactory PyPI virtual repository
  3. Artifactory HuggingFace ML remote repository

• lifecycle_configuration.sh

  Code of the lifecycle configuration script to launch a sagemaker studio lab IDE

sagemaker subdirectory

• Dockerfile

  Dockerfile to build a custom image for inference

  Example docker commands to log in to the Artifactory Docker registry and then to build and push the inference image to Artifactory:

  docker login myartifactory.jfrog.io -u <USER> -p <TOKEN>
  
  docker build -t \
    myartifactory.jfrog.io/sagemaker-docker-virtual/sagemaker/inference-service:1.0_huggingface \
    --build-arg "ARTIFACTORY_DOCKER_REGISTRY=myartifactory.jfrog.io" \
    --build-arg "ARTIFACTORY_DOCKER_REPO=sagemaker-docker-virtual" .
  
  docker push myartifactory.jfrog.io/sagemaker-docker-virtual/sagemaker/inference-service:1.0_huggingface
  

• entrypoint.py

  Entrypoint script used in the custom Docker image to set up the inference handler service

• inference.py

  Code included in the custom Docker image to handle inference using a custom Hugging Face ML model

• deploy-model.py

  Code to deploy the trained model as a SageMaker real-time inference

• test-inference.py

  Code to test the deployed SageMaker real-time inference

• deploy-lambda.sh
• deploy-secret.sh
• lambda_function.py

• Dockerfile

  Dockerfile to build a model train image

  Example docker commands to log in to the Artifactory Docker registry and then to build and push the training image to Artifactory:

  docker login myartifactory.jfrog.io -u <USER> -p <TOKEN>
  
  docker build -t \
    myartifactory.jfrog.io/sagemaker-docker-virtual/sagemaker/train:1.0_huggingface \
    --secret id=pipconfig,src=pip.conf \
    --build-arg "ARTIFACTORY_DOCKER_REGISTRY=myartifactory.jfrog.io" \
    --build-arg "ARTIFACTORY_DOCKER_REPO=sagemaker-docker-virtual" .
  
    docker push myartifactory.jfrog.io/sagemaker-docker-virtual/sagemaker/train:1.0_huggingface
  

• pip.conf

  File used to configure the Artifactory PyPI repository for use during the Python package installation section of the Docker image build (used as a Docker build secret)

• requirements.txt

  Python package requirements for use during the custom Docker image build

  These requirements satisfy everything needed for both the train.py and inference.py scripts (both are combined to simplify the demo)

• run-train-job.py

  Code to create and execute SageMaker train job using a custom Docker image

• train.py

  Model train code that the SageMaker train job will execute