AI Paper Atlas

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

Rethinking Pretraining Data Detection for LLMs: From Local to Global

Chenye Ke, Yan Zhuang, Zirui Liu, Qi Liu

ICML 2026 · Large language models › Language, safety and interpretability · approach: Data and hallucination detection

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
add self-information calibration and volatility divergence to NLL volatility (ablation)AECA · WikiMIA+MIMIR (Pythia-2.8B/6.9B)average AUC: 0.606 -> 0.615Improved
add entropic convolution on top of calibration (ablation)AECA · WikiMIA+MIMIR (Pythia-2.8B/6.9B)average AUC: 0.615 -> 0.619Improved
raise volatility coefficient lambda from 0 to 5 (ablation)AECA · WikiMIA length 128, Pythia-12BAUC: 0.582 -> 0.716Improved
raise volatility coefficient lambda from 0 to 5 (ablation)AECA · WikiMIA length 128, GPT-Neo-2.7BAUC: 0.599 -> 0.698Improved
raise lambda beyond 5 up to 10 (ablation)AECA · WikiMIA length 128, Pythia-12BAUC: 0.716 -> 0.714No clear effect
vary Laplace smoothing alpha from 0.01 to 3 (ablation)AECA · WikiMIA length 128, Pythia-12BAUC: 0.716 and 0.717No clear effect
use small cross-domain Case-law reference corpus instead of large C4 (ablation)AECA · WikiMIA, Pythia-12BAUC: 0.716 -> 0.705Hurt
use larger 10GB C4 reference corpus instead of 1GB (ablation)AECA · WikiMIA, Pythia-12BAUC: 0.711 -> 0.716Improved
replace PPL with AECAGPT-NeoX-20B · WikiMIA length 256 (original)AUC: 0.713 -> 0.758Improved
replace Min-K% with AECAGPT-NeoX-20B · WikiMIA length 256 (original)AUC: 0.743 -> 0.758Improved
replace best baseline with AECAMIMIR average · MIMIR three subsetsaverage AUC: 0.540 -> 0.546Improved
replace Min-K%++ with AECA at length 256WikiMIA length 256 · WikiMIA length 256 (average)average AUC: 0.619 -> 0.714Improved

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