PrE-Text: Training Language Models on Private Federated Data in the Age of LLMs

TL;DR

PrE-Text introduces a method for generating differentially private synthetic textual data that enables training small models more efficiently and improves large language model performance on private data, addressing key challenges of on-device training.

Contribution

The paper presents PrE-Text, a novel approach for DP synthetic data generation that enhances privacy-preserving model training and outperforms on-device training in efficiency and effectiveness.

Findings

PrE-Text synthetic data outperforms on-device training for small models.

Training large models on PrE-Text data improves LLM performance on private data.

PrE-Text reduces communication, computation, and training rounds significantly.

Abstract

On-device training is currently the most common approach for training machine learning (ML) models on private, distributed user data. Despite this, on-device training has several drawbacks: (1) most user devices are too small to train large models on-device, (2) on-device training is communication- and computation-intensive, and (3) on-device training can be difficult to debug and deploy. To address these problems, we propose Private Evolution-Text (PrE-Text), a method for generating differentially private (DP) synthetic textual data. First, we show that across multiple datasets, training small models (models that fit on user devices) with PrE-Text synthetic data outperforms small models trained on-device under practical privacy regimes ($\epsilon=1.29$, $\epsilon=7.58$). We achieve these results while using 9$\times$ fewer rounds, 6$\times$ less client computation per round, and 100$\times$ less communication per round. Second, finetuning large models on PrE-Text's DP synthetic data improves large language model (LLM) performance on private data across the same range of privacy budgets. Altogether, these results suggest that training on DP synthetic data can be a better option than training a model on-device on private distributed data. Code is available at https://github.com/houcharlie/PrE-Text.