English(EN) DP-LAC: Lightweight Adaptive Clipping for Differentially Private Federated Fine-tuning of Language Models

新的DP-LAC方法增强了私有联邦LLM微调

作者 PulseAugur 编辑部 · [1 个来源] · 2026-05-11 09:32

研究人员开发了DP-LAC，一种用于语言模型差分隐私联邦微调的新方法。该技术通过估计初始裁剪阈值并在训练过程中进行调整，而无需额外的隐私成本或新的超参数，从而改进了现有的自适应裁剪方法。与最先进的自适应裁剪和标准DP-SGD方法相比，DP-LAC的平均准确率提高了6.6%。 AI

影响改进了用于协作LLM训练的隐私保护技术，可能实现更安全的设备端模型适应。

排序理由该集群包含一篇学术论文，详细介绍了用于语言模型差分隐私联邦微调的新方法。[lever_c_demoted from research: ic=1 ai=1.0]

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Mete Ozay · 2026-05-11 09:32

DP-LAC: Lightweight Adaptive Clipping for Differentially Private Federated Fine-tuning of Language Models

Federated learning (FL) enables the collaborative training of large-scale language models (LLMs) across edge devices while keeping user data on-device. However, FL still exposes sensitive information through client-provided gradients. Differentially private stochastic gradient de…