Rubric Judgments Lack Uncertainty: Conformal Weighting for LLM Evaluation

Key points

• Rubric-based evaluation is widely used to assess large language model (LLM) outputs by decomposing quality into multiple criteria and assigning each a binary judgment from an LLM verifier.1
• As a result, uncertain judgments remain incorporated into reward estimation without distinction, potentially yielding misleading training signals for downstream reinforcement learning (RL).1
• Specifically, URW elicits a soft uncertainty signal from a fixed pool of paraphrased judgment templates and converts it into a per-criterion weight using split-conformal $p$-values with provable coverage guarantees.1
• Extensive experiments demonstrate that URW consistently improves pairwise preference accuracy and delivers stronger downstream RL alignment.1

Sources (1)

• [1]Rubric Judgments Lack Uncertainty: Conformal Weighting for LLM Evaluation
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 15, 04:06 AM

  To address this limitation, we propose Uncertainty-aware Rubric Weighting (URW), a training-free framework that quantifies criterion-level judgment uncertainty via conformal calibration.

  Rubric-based evaluation is widely used to assess large language model (LLM) outputs by decomposing quality into multiple criteria and assigning each a binary judgment from an LLM verifier.

Extractive summary: sentences quoted from the sources.