//04 area
Evaluation and quality
Evals, LLM as judge, test sets, regression in CI, measuring hallucination and drift over time.
evalsLLM-as-judgeregressiondriftpromptfoo
//
All articles
1- Evaluation and quality
Testing a model before trusting it
I measured a decision model across 1,202 documents for six cents. The result matters less than the method, and the method caught me publishing the wrong conclusion.