Testing AI apps With Evaluations

Use evals to find failures and check that changes helped improve the application.

Top Takeaways

Recognize when an AI application needs an eval
Learn the four parts of an eval: dataset, model, grader, and metrics
Build a dataset with both normal and difficult test cases
Explain how a grader decides if an output meets expected behavior
Read metrics by category to see where an app fails and if a change helped

Class Overview

Live Via

zoom

A promising demo can make an AI application feel ready for launch. You try a few questions, the responses look strong, and the team starts picturing how customers will use it. Then real users show up. They phrase requests differently, leave out important details, and run into situations nobody tested. The question is not whether the application can produce one...

Available Class FormatsPeople Count

Lecture Followed by Q&A

Best for employees who like to listen, absorb and ask questions at the end. Can also be a keynote or speech.

Best for teams up to 120 people

Class Durations

Flexible class lengths to meet your team’s needs.

60Mins
75Mins
90Mins
2Hours

Add to Wishlist

Sol F.

View Profile

Contact us for more information about this class, or sign up to see more content like this to book directly.

Free Account Perks

Save your favorite classes
Full video introductions
Explore all classes and instructors
Personalized recommendations

Other Classes by Sol F.

5 Classes