AI Paper Atlas

Research map / Large language models / Alignment and preferences

Nonparametric LLM Evaluation from Preference Data

Dennis Frauen, Athiya Deviyani, Mihaela van der Schaar, Stefan Feuerriegel

ICML 2026 · Large language models › Alignment and preferences · approach: Active and token-level preference learning · 0 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 debiased estimator instead of plug-inBorda score, synthetic · n=1000estimation error: 0.38 -> 0.15Improved
use debiased estimator instead of plug-inBT score, synthetic · n=1000estimation error: 0.62 -> 0.25Improved
use debiased estimator instead of plug-inRank centrality, synthetic · n=3000estimation error: 0.18 -> 0.09Improved
use debiased estimator instead of plug-inBorda score, synthetic · n=1000CI coverage: 0.17 -> 0.94Improved
use debiased estimator instead of plug-inBT score, synthetic · n=2000CI coverage: 0.85 vs 0.07Mixed
use A-optimal labeling policy instead of randomBorda estimator · budget 2000ranking MSE: 0.141 -> 0.130Improved
use A-optimal labeling policy instead of randomBT estimator · budget 2000ranking MSE: 2.974 -> 2.861Improved
use A-optimal labeling policy instead of randomRank centrality estimator · budget 2000ranking MSE: 0.020 -> 0.017Improved
add higher quality external judges as featuresdebiased estimators · synthetic judgesrelative ranking errorImproved
use debiased BT projection instead of restricted debiased estimator under BT misspecificationDMLRank BT · BT misspecificationestimation 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