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.

翻译：设备端训练是目前在私有、分布式用户数据上训练机器学习模型最常用的方法。尽管如此，设备端训练存在若干缺点：(1) 大多数用户设备容量过小，无法在设备端训练大型模型；(2) 设备端训练对通信和计算资源要求高；(3) 设备端训练难以调试和部署。为解决这些问题，我们提出了私有进化文本方法，一种用于生成差分隐私合成文本数据的方法。首先，我们证明在多个数据集上，使用PrE-Text合成数据训练的小型模型（可适配用户设备的模型）在实用隐私预算下（$\epsilon=1.29$，$\epsilon=7.58$）的表现优于在设备端训练的小型模型。我们在实现这些结果的同时，将训练轮数减少了9倍，每轮客户端计算量减少了6倍，每轮通信量减少了100倍。其次，在PrE-Text生成的差分隐私合成数据上对大模型进行微调，可在相同隐私预算范围内提升大语言模型在私有数据上的性能。综上所述，这些结果表明，在差分隐私合成数据上进行训练可能是比在设备端利用私有分布式数据训练模型更优的选择。代码发布于 https://github.com/houcharlie/PrE-Text。