latent.agents.judges.correctness

CorrectnessJudge -- scores answer correctness against a reference answer.

Classes

CorrectnessScore

CorrectnessScore()

Score schema for answer correctness against a reference.

Functions

CorrectnessJudge

CorrectnessJudge(model: str, kwargs = {}) -> Judge[CorrectnessScore]

Create a correctness judge that scores responses against reference answers.

Args: model: LiteLLM model identifier. **kwargs: Additional arguments passed to Judge constructor.

Returns: A Judge instance that evaluates answer correctness on a 1-5 scale.

Attributes

PROMPT_TEMPLATE

SYSTEM_PROMPT