In the previous chapter, you learned when and how to scale training jobs using features such as Pipe Mode and distributed training, as well as alternatives to S3.
In this chapter, we'll conclude our exploration of training techniques. You'll learn how to slash down your training costs with Managed Spot Training, how to squeeze every drop of accuracy from your models with Automatic Model Tuning, and how to crack models open with SageMaker Debugger.
This chapter covers the following topics:
- Optimizing training costs with Managed Spot Training
- Optimizing hyperparameters with Automatic Model Tuning
- Exploring models with SageMaker Debugger