AI Paper Atlas

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

Tab-MIA: A Benchmark Dataset for Membership Inference Attacks on Tabular Data in LLMs

Eyal German, Sagiv Antebi, Daniel Samira, Asaf Shabtai, Yuval Elovici

ICLR 2026 · Large language models › Language, safety and interpretability · approach: LLM unlearning and benchmarks · 2 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
fine-tune for 3 epochs instead of 1 (ablation)Min-K++ 20.0% attack on LLaMA-3.1 8B · WTQ, Line-SeparatedAUROC (%): 61.60 -> 93.60Hurt
fine-tune for 3 epochs instead of 1 (ablation)Min-K++ 20.0% attack on Mistral 7B · WTQ, Line-SeparatedAUROC (%): 69.70 -> 97.70Hurt
fine-tune for 3 epochs instead of 1 (ablation)Min-K++ 20.0% attack on LLaMA-3.1 8B · Adult, Line-SeparatedAUROC (%): 55.10 -> 71.10Hurt
use Min-K++ 20.0% instead of LOSS (PPL)MIA on LLaMA-3.1 8B · WikiSQL, Line-Separated, 3 epochsAUROC (%): 76.50 -> 88.90Improved
use Min-K++ 20.0% instead of Min-K 20.0%MIA on LLaMA-3.1 8B · WikiSQL, Line-Separated, 3 epochsAUROC (%): 78.10 -> 88.90Improved
use larger model (LLaMA-3.1 8B vs LLaMA-3.2 3B)MIA target model · WTQ, Line-Separated, 3 epochsAUROC (%): 89.70 -> 93.60Hurt
encode with Line-Separated instead of JSON (ablation)Mistral 7B · WTQAUROC (%): 82.70 -> 97.70Hurt
encode with Key-Value Pair instead of JSON (ablation)Mistral 7B · California HousingAUROC (%): 54.50 -> 92.60Hurt
encode with Key-Value Pair instead of HTML (ablation)Mistral 7B · California HousingAUROC (%): 50.60 -> 92.60Hurt
attack with mismatched format (Key-Value) vs matched MarkdownGemma-3 4B · TabFact cross-formatAUROC (%): 85.2 -> 68.9Hurt
attack with mismatched format (Line-Separated) vs matched MarkdownGemma-3 4B · TabFact cross-formatAUROC (%): 85.2 -> 69.4Hurt
measure pretrained model with no fine-tuningLlama-3.1 8B, Key-Value Pair · WTQ pretrainedAUROC (%): 72.0Mixed

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