Summary
Built-in frameworks are extremely useful, but sometimes you need something a little—or very—different. Whether starting from built-in containers or from scratch, SageMaker lets you build your training and deployment containers exactly the way you want them. Freedom for all!
In this chapter, you learned how to customize Python and R containers for data processing, training, and deployment. You saw how you could use them with the SageMaker SDK and its usual workflow. You also learned about MLflow, a nice open source tool that lets you train and deploy models using a CLI.
This concludes our extensive coverage of modeling options in SageMaker: built-in algorithms, built-in frameworks, and custom code. In the next chapter, you'll learn about SageMaker features that help you to scale your training jobs.
