EvalKit Docs

    Evaluators, collections and calibration

    Evaluation rules, dry-run test, usage, duplicate, delete and restore; evaluator collections; judge calibration and the threshold sweep.

    This page — Evaluators, collections and calibration — is being written now and will replace this note in the next edit.

    PreviousEvaluation runsNextOnline rules and trace evaluation

    EvalKit is built by Syntropylabs. Published on PyPI and npm.