AI Paper Atlas

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

LORD-GoF: A Robust Online Detection Approach for LLM Watermarks in Sparse and Mixed Streams

Jiade Xu, Zhouping Li

ICML 2026 · Large language models › Language, safety and interpretability · approach: Text watermarking

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
replace fixed threshold alpha=0.05 with LORD online levelsGoF watermark detection · OPT-1.3B Gumbel-Max stream M=10000, pi=0.05FDR: about 0.50 -> near 0.05 targetImproved
replace fixed threshold with LORD in extremely sparse streamGoF watermark detection · Qwen-2.5-3B Inverse Transform, pi=0.01FDR: 0.85 -> 0.00Improved
increase global sparsity pi toward 0.99 (dense stream) (ablation)LORD-GoF · Qwen-2.5-3B Inverse Transformpower: converges to 1.0Improved
use Anderson-Darling statistic at tau=0.5 (ablation)LORD-GoF · Sheared-LLaMA-2.7B Gumbel-Maxpower / FDR: Power 0.854, FDR 0.000Improved
use Tr-GoF (Phi) instead of other GoF tests at tau=0.3 (ablation)LORD-GoF · Sheared-LLaMA-2.7B Gumbel-Maxpower / FDR: Power 0.839, FDR 0.000Improved
use sensitive GoF tests at tau=0.7 (ablation)LORD-GoF · Sheared-LLaMA-2.7B Gumbel-MaxFDR: e.g. LORD-Chi FDR 0.094, LORD-Cra 0.079Hurt
use Phi test (ablation)LORD-GoF · OPT-1.3B Gumbel-Max streampower: Power not above 0.92 (And and Chi above 0.92)Hurt
replace LORD with LORD++ (asymmetric reward) (ablation)LORD-GoF · appendix comparisonpower: absolute power gap at most 0.006No 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