🤷🏻 Edge#211: What to Test in ML Models

In this issue:

• we discuss what to test in ML models;

• we explain how Meta uses A/B testing to improve Facebook’s newsfeed algorithm;

• we explore Meta’s Ax, a framework for A/B testing in PyTorch.   

Enjoy the learning!  

---

💡 ML Concept of the Day: What to Test in ML Models 

Continuing our series about ML testing, in this issue, we would like to focus on the key properties of ML models that should be targeted in the testing workflows. This is an incredibly important point as different ML testing methods are optimized to validate different properties of ML models. Typically, an ML testing workflow should cover a variety of functional and non-functional properties that provide a holistic view of an ML model’s behavior. Some of the following properties are among the fundamental aspects to be tested in ML models: