A golden dataset is a trusted, human-verified set of examples used as the reference standard for evaluation. Because it is verified, it can anchor regression tests and calibrate LLM judges. Keep it small, stable, and representative rather than large and noisy.
Why it matters
A verified reference set is what lets you trust a quality score. It anchors regression tests and serves as the calibration target for LLM judges.
How it works
Examples are verified by humans and kept small, stable, and representative rather than large and noisy. When scores drop against the golden set, you know the change - not the data - caused it.