Skip to main content

module cleanlab_codex.types.response_validation

Types for response validation.

class BaseResponseValidationResult​

class SingleResponseValidationResult​

Result of a single response validation check.

This class represents the outcome of an individual validation check performed on an AI response.

class SingleResponseValidationResult(BaseModel):
     name: 'typing.Literal['fallback', 'untrustworthy', 'unhelpful']'
     fails_check: 'bool'
     score: 'typing.Dict[str, float]'
     metadata: 'typing.Dict[str, typing.Any]'

Args:

  • name (typing.Literal[‘fallback’, ‘untrustworthy’, ‘unhelpful’]): The name of the validation check.
  • fails_check (bool): Whether the check failed. True if the check failed, False otherwise.
  • score (typing.Dict[str, float]): The score of the response. Typically a single score’s value is between 0.0 and 1.0, but this can vary by check.
  • metadata (typing.Dict[str, typing.Any]): Additional metadata about the response. This can include the threshold values, or other information relevant to the check.

class AggregatedResponseValidationResult​

Result of multiple combined response validation checks.

This class aggregates multiple SingleResponseValidationResults and provides a combined validation outcome.

The class is typically used in a boolean context to determine if any of the underlying checks failed. But each of the individual results are also accessible, via the results field.

class AggregatedResponseValidationResult(BaseModel):
     name: 'typing.Literal['bad']'
     results: 'typing.List[cleanlab_codex.types.response_validation.SingleResponseValidationResult]'

Args:

  • name (typing.Literal[‘bad’]): The name of the aggregated validation check.
  • results (typing.List[cleanlab_codex.types.response_validation.SingleResponseValidationResult]): The individual results of the validation checks.

property fails_check​

Whether any of the underlying checks failed.