## Amazon SageMaker Optuna HPO Blog

This sample code demonstrates how to build an Amazon SageMaker environment for HPO using [Optuna](https://github.com/pfnet/optuna) (an open source hyperparameter tuning framework).

You can set up the environment depicted below with the [CloudFormation template](https://github.com/aws-samples/aws-sagemaker-optuna-hpo-blog/blob/master/template/optuna-template.yaml).

[![Launch Stack](https://s3.amazonaws.com/cloudformation-examples/cloudformation-launch-stack.png)](https://console.aws.amazon.com/cloudformation/home?region=us-east-1#/stacks/new?stackName=optuna-blog&templateURL=https://ml-blog.s3.ap-northeast-1.amazonaws.com/optuna/optuna-template.yaml) 

<div align="center"><img src="images/optuna-architecture.png" width="480"/></div>

Then clone this repository and try examples in the `examples/` directory. The blog post is basically structured to demonstrate the `examples/pytorch_simple/`.

## License Summary

This sample code is made available under a modified MIT license. See the LICENSE file.