Privacy Enhanced PEFT: Tensor Train Decomposition Improves Privacy Utility Tradeoffs under DP-SGD

TL;DR

This paper introduces TTLoRA, a structurally constrained PEFT method that enhances privacy-utility tradeoffs in language model fine-tuning under differential privacy, by shrinking parameter space while maintaining expressivity.

Contribution

The paper proposes TTLoRA and TTLoRA-DP, novel privacy-preserving PEFT architectures with efficient DP-SGD implementation, improving privacy protection and utility over existing methods.

Findings

TTLoRA-DP outperforms LoRA-DP in privacy protection.

TTLoRA uses significantly fewer parameters than LoRA.

TTLoRA exhibits lower membership leakage even without DP.

Abstract

Fine-tuning large language models on sensitive data poses significant privacy risks, as membership inference attacks can reveal whether individual records were used during training. While Differential Privacy (DP) provides formal protection, applying DP to conventional Parameter-Efficient Fine-Tuning (PEFT) methods such as Low-Rank Adaptation (LoRA) often incurs substantial utility loss. In this work, we show that a more structurally constrained PEFT architecture, Tensor Train Low-Rank Adaptation (TTLoRA), can improve the privacy-utility tradeoff by shrinking the effective parameter space while preserving expressivity. To this end, we develop TTLoRA-DP, a differentially private training framework for TTLoRA. Specifically, we extend the ghost clipping algorithm to Tensor Train cores via cached contraction states, enabling efficient Differentially Private Stochastic Gradient Descent (DP-SGD) with exact per-example gradient norm computation without materializing full per-example gradients. Experiments on GPT-2 fine-tuning over the Enron and Penn Treebank datasets show that TTLoRA-DP consistently strengthens privacy protection relative to LoRA-DP while maintaining comparable or better downstream utility. Moreover, TTLoRA exhibits lower membership leakage even without DP training, using substantially smaller adapters and requiring on average 7.6X fewer parameters than LoRA. Overall, our results demonstrate that TTLoRA offers a practical path to improving the privacy-utility tradeoff in parameter-efficient language model adaptation.