用同态加密保护私有数据微调大模型，无需昂贵显卡
Private LoRA Fine-tuning of Open-Source LLMs with Homomorphic Encryption
- 结合LoRA与同态加密，远程计算时保护数据和梯度隐私
- 成功微调Llama-3.2-1B模型，支持HE量化并实测GPU性能
- 适合医疗、法律等敏感数据场景，客户端只需低配设备
在敏感应用中保护开源大语言模型（LLM）微调过程中的数据机密性至关重要。本文提出一种交互式协议，将低秩适配（LoRA）技术用于私有微调。同态加密（HE）保护远程计算节点处理的训练数据和梯度，这些节点负责涉及基础模型权重的大量计算。数据所有者仅需少量本地算力和内存即可协调训练，避免对昂贵客户端显卡的需求。我们通过微调Llama-3.2-1B模型验证了可行性，展示了兼容HE的量化方案的收敛结果，并提供了在GPU硬件上的HE计算性能基准。该方法适用于保密知识库问答、私有代码库微调以构建AI编程助手、基于公司邮件存档生成邮件的AI代理，以及分析敏感法律或医疗文档的模型适配。
原文摘要 · Abstract (English)
Preserving data confidentiality during the fine-tuning of open-source Large Language Models (LLMs) is crucial for sensitive applications. This work introduces an interactive protocol adapting the Low-Rank Adaptation (LoRA) technique for private fine-tuning. Homomorphic Encryption (HE) protects the confidentiality of training data and gradients handled by remote worker nodes performing the bulk of computations involving the base model weights. The data owner orchestrates training, requiring minimal local computing power and memory, thus alleviating the need for expensive client-side GPUs. We demonstrate feasibility by fine-tuning a Llama-3.2-1B model, presenting convergence results using HE-compatible quantization and performance benchmarks for HE computations on GPU hardware. This approach enables applications such as confidential knowledge base question answering, private codebase fine-tuning for AI code assistants, AI agents for drafting emails based on a company's email archive, and adapting models to analyze sensitive legal or healthcare documents.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。