Posted On: Dec 3, 2021
SageMaker Model Registry, a purpose-built service which enables customers to catalog their ML models, now provides endpoint visibility from Studio UI, ability to store custom metadata and view/store broad array of metrics for a given model.
SageMaker Model Registry catalogs customer’s models in a logical group (a.k.a. model group) and stores incremental versions of models as model package versions. Now, customers can associate custom metadata and custom metrics on a model package version. They can also store a broad array of metrics and baselines on a model package version; such as- data quality, model quality, model bias and model explainability. In addition, customers can also view the SageMaker inference endpoint where the model is hosted and dive deep into endpoint metadata.