AWS · MLA-C02 · BETA
Training jobs and hyperparameter tuning
MLA-C02 · Lesson 9 of 23 · 1:08:38
YouTube loads only after you press play. This chapter opens at its timestamp in the full course. Watch on YouTube ↗
In this lesson
Lesson 9 of the free MLA-C02 beta series (AWS Certified Machine Learning Engineer – Associate, domain 2, model development). In this lesson: script mode and ModelTrainer, automatic model tuning with early stopping and warm start, distributed training and Managed Spot Training, and epochs, steps, batch size, and learning rate. With three quick checks and a recap, and AnyCompany Mobility, a car sharing service, as the running example.
Preview the topics
- A team tunes six continuous hyperparameters, and each job is expensive: it wants the fewest jobs. Each new job should learn from the previous results. Which tuning strategy fits best?
- A training job runs nine hours each week, and the team wants the lowest cost. A delay of a few hours is fine, but the job must not restart from scratch after an interruption. What should the ML engineer do?
- The loss swings up and down from one step to the next, and after many epochs, it never settles. The data is already normalized. Which change should the ML engineer try first?