Microsoft · AI-300
Evaluations
AI-300 · Lesson 12 of 15 · 1:26:37
YouTube loads only after you press play. This chapter opens at its timestamp in the full course. Watch on YouTube ↗
In this lesson
Prepare for the AI-300 exam (Microsoft Certified: Machine Learning Operations Engineer Associate). In this lesson: evaluation datasets and data mapping, quality evaluators like groundedness and relevance, risk and safety evaluators, cloud evaluations, GitHub Actions on pull requests, and shadow rating. Includes three original practice questions.
Preview the topics
- In a cloud evaluation, groundedness fails on every row, while relevance works. The dataset has the fields query, response, and Context, with a capital C. What is the most likely cause?
- A safety evaluation gives one answer a violence score of 6. How should the team read this result?
- A team wants automated groundedness scores to block deployments. Reviewers also scored 150 answers by hand. What should the team check first?