domains / data / llm-judge-calibration / validate-evaluator

validate-evaluator

Skill card

authentic

The two copies are nearly identical in structure, steps, metrics, and code. The ai-evals-course copy is the more current, non-archived repository and includes a small but real difference: it references the companion skill `write-code-eval` for code-based evaluators, whereas the archived hamelsmu copy says to use standard unit tests. The ai-evals-course copy is the procedure the other reproduces, so it is the authentic source.

Repositories

Other copies