Skip to content

Glossary: Evaluating LLM Apps

Harness

Also written: eval harness

The code that runs an app on every test case and saves the outputs. A fuller harness can also grade the outputs and produce a report.

Where the book teaches it

Related terms

EvalGrader