# Copyright (c) Microsoft Corporation.
# Licensed under the MIT license.
import enum
from typing import Optional
import yaml
from pathlib import Path
from pyrit.common.path import DATASETS_PATH
from pyrit.models import PromptRequestPiece, SeedPrompt
from pyrit.models.score import UnvalidatedScore, Score
from pyrit.prompt_target import PromptChatTarget
from pyrit.score.scorer import Scorer
TRUE_FALSE_QUESTIONS_PATH = Path(DATASETS_PATH, "score", "true_false_question").resolve()
[docs]
class TrueFalseQuestionPaths(enum.Enum):
CURRENT_EVENTS = Path(TRUE_FALSE_QUESTIONS_PATH, "current_events.yaml").resolve()
GROUNDED = Path(TRUE_FALSE_QUESTIONS_PATH, "grounded.yaml").resolve()
PROMPT_INJECTION = Path(TRUE_FALSE_QUESTIONS_PATH, "prompt_injection.yaml").resolve()
QUESTION_ANSWERING = Path(TRUE_FALSE_QUESTIONS_PATH, "question_answering.yaml").resolve()
GANDALF = Path(TRUE_FALSE_QUESTIONS_PATH, "gandalf.yaml").resolve()
[docs]
class TrueFalseQuestion:
"""
A class that represents a true/false question.
This is sent to an LLM and can be used as an alternative to a yaml file from TrueFalseQuestionPaths.
"""
[docs]
def __init__(
self, *, true_description: str, false_description: str = "", category: str = "", metadata: Optional[str] = ""
):
self.true_description = true_description
self.false_description = (
false_description or "Anything that does not fulfill or is not related to the true description."
)
self.category = category
self.metadata = metadata
self._keys = ["category", "true_description", "false_description"]
def __getitem__(self, key):
return getattr(self, key)
def __setitem__(self, key, value):
setattr(self, key, value)
def __iter__(self):
# Define which keys should be included when iterating
return iter(self._keys)
[docs]
class SelfAskTrueFalseScorer(Scorer):
"""A class that represents a self-ask true/false for scoring."""
[docs]
def __init__(
self,
*,
chat_target: PromptChatTarget,
true_false_question_path: Optional[Path] = None,
true_false_question: Optional[TrueFalseQuestion] = None,
true_false_system_prompt_path: Optional[Path] = None,
) -> None:
self._prompt_target = chat_target
self.scorer_type = "true_false"
if not true_false_question_path and not true_false_question:
raise ValueError("Either true_false_question_path or true_false_question must be provided.")
if true_false_question_path and true_false_question:
raise ValueError("Only one of true_false_question_path or true_false_question should be provided.")
if true_false_question_path:
true_false_question = yaml.safe_load(true_false_question_path.read_text(encoding="utf-8"))
for key in ["category", "true_description", "false_description"]:
if key not in true_false_question:
raise ValueError(f"{key} must be provided in true_false_question.")
self._score_category = true_false_question["category"]
true_category = true_false_question["true_description"]
false_category = true_false_question["false_description"]
metadata = true_false_question["metadata"] if "metadata" in true_false_question else ""
true_false_system_prompt_path = (
true_false_system_prompt_path
if true_false_system_prompt_path
else TRUE_FALSE_QUESTIONS_PATH / "true_false_system_prompt.yaml"
)
scoring_instructions_template = SeedPrompt.from_yaml_file(true_false_system_prompt_path)
self._system_prompt = scoring_instructions_template.render_template_value(
true_description=true_category, false_description=false_category, metadata=metadata
)
[docs]
async def score_async(self, request_response: PromptRequestPiece, *, task: Optional[str] = None) -> list[Score]:
"""
Scores the given request_response using "self-ask" for the chat target and adds score to memory.
Args:
request_response (PromptRequestPiece): The prompt request piece containing the text to be scored.
task (str): The task based on which the text should be scored (the original attacker model's objective).
Currently not supported for this scorer.
Returns:
list[Score]: The request_response scored.
The category is configured from the TrueFalseQuestionPath.
The score_value is True or False based on which fits best.
metadata can be configured to provide additional information.
"""
self.validate(request_response, task=task)
unvalidated_score: UnvalidatedScore = await self._score_value_with_llm(
prompt_target=self._prompt_target,
system_prompt=self._system_prompt,
prompt_request_value=request_response.converted_value,
prompt_request_data_type=request_response.converted_value_data_type,
scored_prompt_id=request_response.id,
category=self._score_category,
task=task,
)
score = unvalidated_score.to_score(score_value=unvalidated_score.raw_score_value)
self._memory.add_scores_to_memory(scores=[score])
return [score]
[docs]
def validate(self, request_response: PromptRequestPiece, *, task: Optional[str] = None):
if task:
raise ValueError("This scorer does not support tasks")
