Evalkeeper
    Preparing search index...

    Function attackMistakes

    • A model's serious mistakes on the attacks: its serious mistakes on every injection suite, added up. Null when any of them has no current, complete test of it, so it can't be compared yet; also null with no injection suites.

      Parameters

      • key: string

        The model's row key (model and reasoning setting).

      • attacks: readonly AttackResults[]

        The feature's injection suites.

      Returns number | null