ContinuousBench shows non-private synthetic text transfers corpus-specific capabilities while state-of-the-art DP methods fail to do so even at ε=100.
Georgi Ganev, Meenatchi Sundaram Muthu Selva Annamalai, Samir Mahiou, and Emiliano De Cristofaro
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
PersonaLedger LLM simulator achieves AUC 0.70 for fraud detection at epsilon=1 from DP inputs but shows significant distribution drift due to learned priors overriding input statistics on temporal and demographic features.
citing papers explorer
-
ContinuousBench: Can Differentially Private Synthetic Text Improve Capabilities?
ContinuousBench shows non-private synthetic text transfers corpus-specific capabilities while state-of-the-art DP methods fail to do so even at ε=100.
-
Evaluating LLM Simulators as Differentially Private Data Generators
PersonaLedger LLM simulator achieves AUC 0.70 for fraud detection at epsilon=1 from DP inputs but shows significant distribution drift due to learned priors overriding input statistics on temporal and demographic features.