AI ENGVisual Encyclopedia

MODULE 5: LLM-AS-JUDGE · SCENE 13

Rubric Design & Pointwise Scoring

Criteria, anchors, and output schema. A judge is only as good as the rubric that tells it what a 3 means versus a 4.

RUBRIC NOISE σ0.5JUDGE NOISE σ0.5REPRODUCIBLE · 67%
OBSERVED JUDGE VARIANCE = TRUE + RUBRIC + JUDGEMOSTLY SIGNAL — THE RUBRIC EARNS ITS KEEP

Push both noises to zero and the bar turns fully teal — that is what anchored, single-axis, low-temperature rubrics buy.

TECHNICAL BREAKDOWNModule 5: LLM-as-Judge

A judge is only as good as its rubric

When there is no single correct answer, an LLM can grade open-ended output against a rubric. The rubric defines the criteria, the score levels, and observable anchors for each level. Vague rubrics produce noisy, self-inconsistent scores that are impossible to audit.

Criteria and anchors

Define each score level by observable properties ('cites a source', 'arithmetic is correct'), not adjectives like 'good'.

One axis at a time

Scoring helpfulness, correctness, and tone in one pass compounds errors. Split into separate judgments.

Structured output

Force a schema (criterion, score, rationale) so scores can be parsed, aggregated, and audited.

MATHEMATICAL FORMULATION · RUBRIC CONSISTENCY
σ_score² = σ_true² + σ_rubric² + σ_judge²

Observed judge variance mixes real quality differences, rubric ambiguity, and judge sampling. Tightening anchors reduces the middle term; lower temperature reduces the last.

REAL-WORLD PRODUCTION ENGINEERING
  • MT-Bench uses a structured rubric with a 1–10 scale and explicit criteria for each axis.
  • Production LLM evaluation often uses binary pass/fail rubrics, which are more reproducible than wide ordinal scales.