PopQuiz Attack infers LLM training data membership by turning examples into quiz questions and measuring answer accuracy, reaching 0.873 average ROC-AUC across six models and outperforming prior methods by 20.6%.
Datasets for large language models: A comprehensive survey
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
A review of 114 studies creates taxonomies for code and data quality issues, formalizes 18 propagation mechanisms from training data defects to LLM-generated code defects, and synthesizes detection and mitigation techniques.
DynaHug trains an OCSVM on dynamic runtime behaviors of benign PTMs and achieves up to 44% higher F1-score than static, dynamic, and LLM-based baselines on over 25,000 models.
Kernel-based distributional discrepancy enables auditing of upstream training data in distilled one-step diffusion models by detecting preserved distributional alignment rather than per-instance memorization.
STELLAR-E modifies the TGRT Self-Instruct framework to produce tailored synthetic LLM evaluation datasets that score an average 5.7% higher on LLM-as-a-judge metrics than existing language-specific benchmarks.
citing papers explorer
-
Pop Quiz Attack: Black-box Membership Inference Attacks Against Large Language Models
PopQuiz Attack infers LLM training data membership by turning examples into quiz questions and measuring answer accuracy, reaching 0.873 average ROC-AUC across six models and outperforming prior methods by 20.6%.
-
Bridging Generation and Training: A Systematic Review of Quality Issues in LLMs for Code
A review of 114 studies creates taxonomies for code and data quality issues, formalizes 18 propagation mechanisms from training data defects to LLM-generated code defects, and synthesizes detection and mitigation techniques.
-
Malicious ML Model Detection by Learning Dynamic Behaviors
DynaHug trains an OCSVM on dynamic runtime behaviors of benign PTMs and achieves up to 44% higher F1-score than static, dynamic, and LLM-based baselines on over 25,000 models.
-
Distributional Statistics Restore Training Data Auditability in One-step Distilled Diffusion Models
Kernel-based distributional discrepancy enables auditing of upstream training data in distilled one-step diffusion models by detecting preserved distributional alignment rather than per-instance memorization.
-
STELLAR-E: a Synthetic, Tailored, End-to-end LLM Application Rigorous Evaluator
STELLAR-E modifies the TGRT Self-Instruct framework to produce tailored synthetic LLM evaluation datasets that score an average 5.7% higher on LLM-as-a-judge metrics than existing language-specific benchmarks.