Summary
In this chapter, we covered model registries and the benefits of utilizing a model registry to manage Amazon SageMaker models at scale. Common patterns for model registry implementations were covered, including Amazon SageMaker's model registry, building a custom model registry using AWS services, and utilizing a third-party or OSS model registry implementation. Each option is a valid choice depending on your use case and needs. However, we also highlighted some of the considerations when choosing the implementation that best fits your requirements.
Finally, we did a deep dive into Amazon SageMaker's model registry, covering detailed recommendations for creating model package groups, as well as registering models by creating model packages.
In the next chapter, we'll cover performing live tests and updates of production models using Amazon SageMaker endpoint production variants.
