Reproduce “LLM-Attack Suite: A Unified Red-Team Framework for Privacy Evaluation of Split and Distributed LLM Systems” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Georgios Politis, Evangelos Pappas
LLM-Attack Suite makes privacy evaluations of split language models reusable and inspectable by recording attacker access, observations, data, metrics, controls, and decision rules, then replaying captured messages and rescoring predictions. In a prespecified GPT-2 experiment, adding returned gradients raised token recovery from 94.20% to 97.38% and exact-document recovery from 13.71% to 37.77%; this result is limited to the tested setup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.