nemoguardrails.library.jailbreak_detection.actions | NVIDIA NeMo Guardrails Library Developer Guide

Module Contents

Functions

Name	Description
`jailbreak_detection_heuristics`	Checks the user’s prompt to determine if it is attempt to jailbreak the model.
`jailbreak_detection_model`	Uses a trained classifier to determine if a user input is a jailbreak attempt

Data

log

API

nemoguardrails.library.jailbreak_detection.actions.jailbreak_detection_heuristics(
    llm_task_manager: nemoguardrails.llm.taskmanager.LLMTaskManager,
    context: typing.Optional[dict] = None,
    kwargs = {}
) -> bool

async

Checks the user’s prompt to determine if it is attempt to jailbreak the model.

nemoguardrails.library.jailbreak_detection.actions.jailbreak_detection_model(
    llm_task_manager: nemoguardrails.llm.taskmanager.LLMTaskManager,
    context: typing.Optional[dict] = None,
    model_caches: typing.Optional[typing.Dict[str, nemoguardrails.llm.cache.CacheInterface]] = None
) -> bool

async

Uses a trained classifier to determine if a user input is a jailbreak attempt

nemoguardrails.library.jailbreak_detection.actions.log = logging.getLogger(__name__)