DP-MemArc: Differential Privacy Transfer Learning for Memory Efficient Language Models

TL;DR

DP-MemArc is a novel training framework that significantly reduces memory usage and enhances privacy protection during fine-tuning of large language models, enabling more resource-efficient and secure deployment.

Contribution

It introduces a new memory-efficient training method with differential privacy guarantees using side or reversible networks for large language models.

Findings

Achieves approximately 2.5x memory reduction.

Provides robust differential privacy during fine-tuning.

Effective across various task scenarios.

Abstract

Large language models have repeatedly shown outstanding performance across diverse applications. However, deploying these models can inadvertently risk user privacy. The significant memory demands during training pose a major challenge in terms of resource consumption. This substantial size places a heavy load on memory resources, raising considerable practical concerns. In this paper, we introduce DP-MemArc, a novel training framework aimed at reducing the memory costs of large language models while emphasizing the protection of user data privacy. DP-MemArc incorporates side network or reversible network designs to support a variety of differential privacy memory-efficient fine-tuning schemes. Our approach not only achieves about 2.5 times in memory optimization but also ensures robust privacy protection, keeping user data secure and confidential. Extensive experiments have demonstrated that DP-MemArc effectively provides differential privacy-efficient fine-tuning across different task scenarios.