Secure Transformer Inference Protocol

Mu Yuan, Lan Zhang, Xiang-Yang Li
11/14/2023
cs.CRcs.LG

Official code: yuanmu97/secure-transformer-inference (98 stars, MIT, last commit 2024-05-08).

Abstract

Security of model parameters and user data is critical for Transformer-based services, such as ChatGPT. While recent strides in secure two-party protocols have successfully addressed security concerns in serving Transformer models, their adoption is practically infeasible due to the prohibitive cryptographic overheads involved. Drawing insights from our hands-on experience in developing two real-world Transformer-based services, we identify the inherent efficiency bottleneck in the two-party assumption. To overcome this limitation, we propose a novel three-party threat model. Within this framework, we design a semi-symmetric permutation-based protection scheme and present STIP, the first secure Transformer inference protocol without any inference accuracy loss. Experiments on representative Transformer models in real systems show that STIP has practical security and outperforms state-of-the-art secure two-party protocols in efficiency by millions of times.

View on arXivView PDF

Code Implementations(1)

yuanmu97/secure-transformer-inferenceOfficial
MIT

[NDSS 2026] Secure Transformer Inference is a protocol for serving Transformer-based models securely.

9818PythonNov 10, 20232 years agoMIT

Similar Papers

CENTAUR: Bridging the Impossible Trinity of Privacy, Efficiency, and Performance in Privacy-Preserving Transformer Inference

83% match2024

Ditto: Quantization-aware Secure Inference of Transformers upon MPC

83% match1 repo3252024

SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPC

82% match1 repo52024

Unelicitable Backdoors in Language Models via Cryptographic Transformer Circuits

80% match2024

LLMs Can Understand Encrypted Prompt: Towards Privacy-Computing Friendly Transformers

80% match1 repo162023

Encryption-Friendly LLM Architecture

79% match2024

Cite this paper

@article{yuan2023secure,
  title  = {Secure Transformer Inference Protocol},
  author = {Mu Yuan and Lan Zhang and Xiang-Yang Li},
  year   = {2023},
  eprint = {2312.00025},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2312.00025v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2312.00025.svg)](https://codeofpaper.com/paper/2312.00025)

Discussion