Skip to content

cmu-mlsp/aws-sagemaker-remote

 
 

Repository files navigation

aws-sagemaker-remote

Remotely run and track ML research using AWS SageMaker.

  • Standardized command line flags
  • Remotely run scripts with minimal changes
  • Automatically manage AWS resources
  • All code, inputs, outputs, arguments, and settings are tracked in one place
  • Reproducible batch processing jobs to prepare datasets
  • Reproducible training jobs that track hyperparameters and metrics

Track three types of objects in a standard way:

  • Processing jobs consume file inputs and produce file outputs. Useful for data conversion, extraction, etc.
  • Training jobs train models while tracking metrics and hyperparameters.
  • Inference models provide predictions and can be deployed on endpoints. Can be automatically created from and linked to training jobs for tracking purposes or can deploy externally-created models.

Installation

Release

pip install aws-sagemaker-remote

Development

git clone https://github.com/bstriner/aws-sagemaker-remote
cd aws-sagemaker-remote
python setup.py develop

Documentation

View latest documentation at ReadTheDocs

Continuous Integration

View continuous integration at TravisCI

PyPI

View releases on PyPI

GitHub

View source code on GitHub

GitHub tags are automatically released on ReadTheDocs, tested on TravisCI, and deployed to PyPI if successful.

About

Remotely run scripts on AWS SageMaker

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 87.9%
  • Dockerfile 11.0%
  • Other 1.1%