AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Noisy but Valid: Robust Statistical Evaluation of LLMs with Imperfect Judges

Chen Feng, Minghe Shen, Ananth Balashankar, Carsten Gerner-Beuerle, Miguel R. D. Rodrigues

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Statistical frameworks for LLM evaluation · 9 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use noisy judge-based test instead of direct human-only test with a capable judgeNoisy HT · Jigsaw, Hate Speech, SafeRLHFtype-II errorImproved
use noisy test with a weak judge (low TPR / high FPR)Noisy HT · synthetic and real datasetstype-II errorHurt
add variance-corrected threshold accounting for calibration uncertainty (ablation)Noisy HT · all settingstype-I errorImproved
know judge TPR/FPR exactly (oracle) instead of estimating (ablation)Noisy HT · synthetic and real datasetstype-II errorImproved
use PPI baselines instead of noisy HTNoisy HT · all settingstype-II errorHurt
increase calibration set size n_M from 25 to 100 (ablation)TPR/FPR estimators · Jigsawestimator standard deviationImproved
use federated prompts instead of single judge prompt (ablation)LLM judge · Jigsawjudge TPR/FPR and type-II errorImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy