sagemill 0.0.1

Run SageMaker Job like Papermil

Sagemill lets you to execute your notebook as SageMaker jobs in the similar way to papermill

Overall features

• parameterize notebooks

• execute notebooks as SageMaker Trainig/Processing Job

Installatioin

$ pip install sagemill

Prerequsites

• Python>=3.5

Required IAM policy

• ecr

  • create-repository

  • get-authorization-token

• sts

  • get-caller-identity

• permission to run Sagemaker jobs

  • write to cloudwatch logs

  • access to some S3 buckets

  • start training/proceessing jobs

  • etc.

Example notebooks

These notebook are assumed to be conda_python3 in SageMaker Notebook instance If you run it on different environments, install conda and run pip install sagemaker

• Training job with tensorflow

• Training job with your own docker image

• Processing job with SKLearnProcessor

• Processing job with your own docker image