Summary

In this chapter, we introduced two features integrated into SageMaker Studio—JumpStart and Autopilot—with three ML use cases to demonstrate low-to-no code ML options for ML developers. We learned how to browse JumpStart solutions in the catalog and how to deploy an end-to-end CV solution from JumpStart to detect defects in products. We also deployed and fine-tuned a question-answering model using the DistilRoBERTa Base model from the JumpStart model zoo without any ML coding. With Autopilot, we built a white wine quality prediction model simply by pointing Autopilot to a dataset stored in S3 and starting an Autopilot job – no code necessary. It turned out that Autopilot even outperforms the model created by the original researchers, which may have taken months of research.

With the next chapter, we begin the next part of the book: Production and Operation of Machine Learning with SageMaker Studio. We will learn how we can move from prototyping to production...