Encryption-Friendly LLM Architecture

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Rho, Donghwan, Kim, Taeseong, Park, Minje, Kim, Jung Woo, Chae, Hyunsik, Ryu, Ernest K., Cheon, Jung Hee
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912239647195136
author Rho, Donghwan
Kim, Taeseong
Park, Minje
Kim, Jung Woo
Chae, Hyunsik
Ryu, Ernest K.
Cheon, Jung Hee
author_facet Rho, Donghwan
Kim, Taeseong
Park, Minje
Kim, Jung Woo
Chae, Hyunsik
Ryu, Ernest K.
Cheon, Jung Hee
contents Large language models (LLMs) offer personalized responses based on user interactions, but this use case raises serious privacy concerns. Homomorphic encryption (HE) is a cryptographic protocol supporting arithmetic computations in encrypted states and provides a potential solution for privacy-preserving machine learning (PPML). However, the computational intensity of transformers poses challenges for applying HE to LLMs. In this work, we propose a modified HE-friendly transformer architecture with an emphasis on inference following personalized (private) fine-tuning. Utilizing LoRA fine-tuning and Gaussian kernels, we achieve significant computational speedups -- 6.94x for fine-tuning and 2.3x for inference -- while maintaining performance comparable to plaintext models. Our findings provide a viable proof of concept for offering privacy-preserving LLM services in areas where data protection is crucial. Our code is available on GitHub.
format Preprint
id arxiv_https___arxiv_org_abs_2410_02486
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Encryption-Friendly LLM Architecture
Rho, Donghwan
Kim, Taeseong
Park, Minje
Kim, Jung Woo
Chae, Hyunsik
Ryu, Ernest K.
Cheon, Jung Hee
Cryptography and Security
Machine Learning
Large language models (LLMs) offer personalized responses based on user interactions, but this use case raises serious privacy concerns. Homomorphic encryption (HE) is a cryptographic protocol supporting arithmetic computations in encrypted states and provides a potential solution for privacy-preserving machine learning (PPML). However, the computational intensity of transformers poses challenges for applying HE to LLMs. In this work, we propose a modified HE-friendly transformer architecture with an emphasis on inference following personalized (private) fine-tuning. Utilizing LoRA fine-tuning and Gaussian kernels, we achieve significant computational speedups -- 6.94x for fine-tuning and 2.3x for inference -- while maintaining performance comparable to plaintext models. Our findings provide a viable proof of concept for offering privacy-preserving LLM services in areas where data protection is crucial. Our code is available on GitHub.
title Encryption-Friendly LLM Architecture
topic Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2410.02486