AI Paper Atlas

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

(Token-Level) \textbf{InfoRMIA}: Stronger Membership Inference and Privacy Assessment for LLMs

Jiashu Tao, Reza Shokri

ICLR 2026 · Large language models › Language, safety and interpretability · approach: LLM unlearning and benchmarks

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 RMIA's thresholded population-count score with continuous InfoRMIA information-theoretic scoreRMIA · AG News, |Z|=100TPR@0.1%FPR: 0.00% -> 12.0%Improved
replace RMIA with InfoRMIARMIA · CIFAR-10, |Z|=1000AUC: 0.8229 -> 0.8330Improved
replace RMIA with InfoRMIARMIA · Purchase100, |Z|=1000AUC: 0.5311 -> 0.5754Improved
increase population size |Z| for RMIA from 100 to 1000 (ablation)RMIA · AG NewsAUC: 0.8574 -> 0.8766Improved
increase population size |Z| for InfoRMIA from 100 to 1000 (ablation)InfoRMIA · AG NewsAUC: 0.8784 -> 0.8784No clear effect
use token-level InfoRMIA averaged over tokens instead of sequence-level InfoRMIA (ablation)InfoRMIA · finetuned GPT-2 sequence-level MIAAUC: AG News 4 ep 0.945 -> 0.942; ai4privacy 4 ep 0.822 -> 0.804; TPR AG News 4 ep 16.2% -> 20.6%Mixed
aggregate token scores by min-k instead of averaging (ablation)token-level InfoRMIA · MIMIR pretrained PythiaTPR at low FPRHurt
use min-k instead of averaging for aggregation (ablation)token-level InfoRMIA · MIMIR pretrained PythiaAUCImproved
use step-1 Pythia-160M checkpoint as reference model rather than later checkpoints (ablation)token-level InfoRMIA · MIMIR pretrained Pythiamembership inference performanceImproved
evaluate private tokens only vs all tokens at sequence levelInfoRMIA on ai4privacy · ai4privacy fine-tuned GPT-2average membership scoreHurt

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