pyrit.score.AzureContentFilterScorer

class AzureContentFilterScorer(*, endpoint: str = None, api_key: str = None, use_aad_auth: bool = False, harm_categories: list[TextCategory] = None)

Bases: Scorer

__init__(*, endpoint: str = None, api_key: str = None, use_aad_auth: bool = False, harm_categories: list[TextCategory] = None) → None

Class that initializes an Azure Content Filter Scorer

Parameters:

• api_key (str, Optional) – The API key for accessing the Azure OpenAI service. Defaults to the API_KEY_ENVIRONMENT_VARIABLE environment variable.

• endpoint (str, Optional) – The endpoint URL for the Azure OpenAI service. Defaults to the ENDPOINT_URI_ENVIRONMENT_VARIABLE environment variable.

• use_aad_auth (bool, Optional) – Attempt to use DefaultAzureCredential If set to true, attempt to use DefaultAzureCredential for auth

• harm_categories – The harm categories you want to query for as per defined in azure.ai.contentsafety.models.TextCategory.

Methods

__init__(*[, endpoint, api_key, ...])	Class that initializes an Azure Content Filter Scorer
get_identifier()	Returns an identifier dictionary for the scorer.
scale_value_float(value, min_value, max_value)	Scales a value from 0 to 1 based on the given min and max values.
score_async(request_response, *[, task])	Evaluating the input text or image using the Azure Content Filter API
score_image_async(image_path, *[, task])	Scores the given image using the chat target.
score_prompts_batch_async(*, request_responses)
score_text_async(text, *[, task])	Scores the given text based on the task using the chat target.
validate(request_response, *[, task])	Validates the request_response piece to score.

Attributes

API_KEY_ENVIRONMENT_VARIABLE
ENDPOINT_URI_ENVIRONMENT_VARIABLE
scorer_type

API_KEY_ENVIRONMENT_VARIABLE: str = 'AZURE_CONTENT_SAFETY_API_KEY'

ENDPOINT_URI_ENVIRONMENT_VARIABLE: str = 'AZURE_CONTENT_SAFETY_API_ENDPOINT'

async score_async(request_response: PromptRequestPiece, *, task: str | None = None) → list[Score]

Evaluating the input text or image using the Azure Content Filter API

Parameters:

• request_response (PromptRequestPiece) – The prompt request piece containing the text to be scored. Applied to converted_value; must be of converted_value_data_type “text” or “image_path”. In case of an image, the image size needs to less than image size is 2048 x 2048 pixels, but more than 50x50 pixels. The data size should not exceed exceed 4 MB. Image must be of type JPEG, PNG, GIF, BMP, TIFF, or WEBP.

• 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:

A Score object with the score value mapping to severity utilizing the get_azure_severity function. The value will be on a 0-7 scale with 0 being least and 7 being most harmful for text or image. Definition of the severity levels can be found at https://learn.microsoft.com/en-us/azure/ai-services/content-safety/concepts/harm-categories? tabs=definitions#severity-levels

Raises:

ValueError if converted_value_data_type is not "text" or "image_path" or image isn't in supported format –

validate(request_response: PromptRequestPiece, *, task: str | None = None)

Validates the request_response piece to score. Because some scorers may require specific PromptRequestPiece types or values.

Parameters:

• request_response (PromptRequestPiece) – The request response to be validated.

• task (str) – The task based on which the text should be scored (the original attacker model’s objective).