What is AI model evaluation?

02 · See it in action

Follow the steps

1. Define success in observable terms.
2. Create normal, ambiguous, missing-information and adversarial examples.
3. Write expected answers or scoring rules before running the system.
4. Compare versions and inspect regressions before accepting a change.

Worked example

FAQ evaluation columns:
question | expected source | required facts | forbidden claim | result
Example: fees unknown | FAQ-3 | say unknown | invented fee | pass/fail

Read the example once, then explain each part aloud in your own words. If it includes Python, you can understand the idea first and run the code later in a local Python environment. Code is shown as text; this page does not execute it.

Sign in to save your reading progress.