pyrit.executor.attack.AttackScoringConfig#

class AttackScoringConfig(objective_scorer: ~pyrit.score.scorer.Scorer | None = None, refusal_scorer: ~pyrit.score.scorer.Scorer | None = None, auxiliary_scorers: ~typing.List[~pyrit.score.scorer.Scorer] = <factory>, use_score_as_feedback: bool = True, successful_objective_threshold: float = 0.8)[source]#

Bases: object

Scoring configuration for evaluating attack effectiveness.

This class defines the scoring components used to evaluate attack effectiveness, detect refusals, and perform auxiliary scoring operations.

__init__(objective_scorer: ~pyrit.score.scorer.Scorer | None = None, refusal_scorer: ~pyrit.score.scorer.Scorer | None = None, auxiliary_scorers: ~typing.List[~pyrit.score.scorer.Scorer] = <factory>, use_score_as_feedback: bool = True, successful_objective_threshold: float = 0.8) None#

Methods

__init__([objective_scorer, refusal_scorer, ...])

Attributes

auxiliary_scorers: List[Scorer]#
objective_scorer: Scorer | None = None#
refusal_scorer: Scorer | None = None#
successful_objective_threshold: float = 0.8#
use_score_as_feedback: bool = True#