luml.experiments.evaluation.scorers.builtin.relevancy

Relevancy Objects

class Relevancy(LLMJudgeScorer)

Scores how relevant a response is to the input question.

Unsupervised scorer — no expected output required. Returns a float 0.0 (completely irrelevant) to 1.0 (fully relevant).

Default input keys: "question", "query".

from luml.experiments.evaluation.scorers.builtin import Relevancy
scorer = Relevancy()
scorer = Relevancy(input_key="prompt")