AI Paper Atlas

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

An Ensemble Framework for Unbiased Language Model Watermarking

Yihan Wu, Ruibo Chen, Georgios Milis, Heng Huang

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Text watermarking · 5 citations (OpenAlex)

Read the paper

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
add ensemble of n=5 keysDiPmark (alpha=0.3) · Detectability, 250 tokensTPR@0.01% FPR: 32.22% -> 66.77%Improved
add ensemble of n=5 keysgamma-reweight · Detectability, 500 tokensTPR@0.01% FPR: 72.45% -> 88.58%Improved
add ensemble of n=3 keysMCMark (l=20) · Detectability, 250 tokensTPR@0.01% FPR: 90.37% -> 91.71%Improved
add ensemble of n=3 keysMCMark (l=20) · Detectability, 500 tokensmedian p-value: 8.30e-26 -> 2.58e-31Improved
increase ensemble size from 5 to 10 (ablation)ENS-gamma-reweight · Detectability, 250 tokensTPR@0.01% FPR: 64.14% -> 55.54%Hurt
increase ensemble size from 5 to 10 (ablation)ENS-DiPmark (alpha=0.4) · Detectability, 250 tokensTPR@0.01% FPR: 67.15% -> 58.74%Hurt
increase ensemble size from 3 to 5 (ablation)ENS-MCMark (l=20) · Detectability, 500 tokensTPR@0.01% FPR: 99.57% -> 98.90%Hurt
vary ensemble size n=1..10 (ablation)DiPmark and gamma-reweight · C4 subset, Llama-3.2-3B-InstructTPR@0.01% FPR and median p-valueMixed
increase generation length (ablation)all watermarking methods · C4 subset, Llama-3.2-3B-InstructTPR@0.01% FPR and median p-value: 250 -> 500 tokens, e.g. 32.22% -> 61.68% (DiPmark)Improved
add ensemble n=3 under GPT-4o-mini paraphrase attackSynthID (m=30) · GPT-4o-mini paraphrase attackTPR@0.01% FPR: 13.47% -> 29.44%Improved
compare ENS-MCMark vs SynthID under back translationSynthID (m=30) · Back translation (En-Fr)TPR@0.01% FPR: 64.53% -> 76.43%Improved
add ensemble to MCMark (n=3)MCMark · WMT En-Ro translationBLEU: 20.35 -> 20.43No clear effect

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