Reproduce “Perseus: A Full-Stack System for Homomorphic Model Inference” 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

Alessandro Zirilli, Davide Marincione, Evgenios M. Kornaropoulos, Giuseppe Ateniese, Emanuele Rodolà

Perseus runs language models on encrypted inputs by combining a cryptographic runtime, data packing, and bootstrap planning. It regenerates key material, uses complex ciphertext slots, and plans bootstrap placement. For Llama-3-8B decoding, it reports 7.8 seconds per token at context 512 and perplexity 7.77 versus 7.76 plaintext. Its reported 3.96× speedup over Sylph compares different contexts and timing protocols.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.