luml.experiments.evaluation.scorers.builtin.completeness

Completeness Objects

class Completeness(LLMJudgeScorer)

Scores whether a response fully addresses all parts of a question or task.

Unsupervised scorer — no expected output required. Returns a float 0.0 (does not address the question) to 1.0 (fully and completely addresses all parts).

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

from luml.experiments.evaluation.scorers.builtin import Completeness
scorer = Completeness()