mirror of
https://github.com/hanzoai/papers.git
synced 2026-10-30 14:58:42 +00:00
zeekay
f9db4f5737
Bylines, paper authorship, correspondence addresses and the founder blurb under each post named an individual. They name the organisation that published the work instead: the brand on posts and pages, the research byline on papers, and research@ in place of a personal address. Both are what these repositories already used elsewhere — nothing new was coined. Co-authors are untouched. Where a paper named someone else alongside, they stay named; only the one byline changed. Bibliography entries, running heads, PDF metadata and affiliations move with it, so a citation still resolves to whoever the cited paper names. Sample and fixture data uses the placeholder name these tests already use elsewhere, so no assertion changed meaning.
68 lines
3.2 KiB
TeX
68 lines
3.2 KiB
TeX
% Hanzo ACI: AI Chain Infrastructure for Decentralized Compute Markets
|
|
\documentclass[11pt,twocolumn]{article}
|
|
\usepackage{../shared/hanzocover}
|
|
\usepackage{amsmath, amssymb, amsthm}
|
|
\usepackage{mathtools}
|
|
\usepackage{bm}
|
|
\usepackage{graphicx}
|
|
\usepackage{booktabs}
|
|
\usepackage{hyperref}
|
|
\usepackage{enumitem}
|
|
\usepackage{algorithm}
|
|
\usepackage{algpseudocode}
|
|
\usepackage{xcolor}
|
|
\usepackage{float}
|
|
\usepackage{subcaption}
|
|
\usepackage{multirow}
|
|
\usepackage{listings}
|
|
\input{../shared/lstlang}
|
|
\usepackage[numbers]{natbib}
|
|
\hypersetup{colorlinks=true,linkcolor=black,citecolor=blue,urlcolor=blue}
|
|
|
|
\newtheorem{theorem}{Theorem}
|
|
\newtheorem{lemma}[theorem]{Lemma}
|
|
\newtheorem{definition}{Definition}
|
|
\newtheorem{proposition}[theorem]{Proposition}
|
|
\newtheorem{corollary}[theorem]{Corollary}
|
|
|
|
\lstset{
|
|
basicstyle=\ttfamily\small,
|
|
breaklines=true,
|
|
frame=single,
|
|
numbers=left,
|
|
numberstyle=\tiny,
|
|
xleftmargin=2em,
|
|
}
|
|
|
|
\title{Hanzo ACI: AI Chain Infrastructure for Decentralized Compute Markets}
|
|
\author{
|
|
Antje Worring, Hanzo AI Research \quad
|
|
David Wei \quad
|
|
Marcus Chen \\
|
|
\textit{Hanzo AI Research} \\
|
|
\texttt{[email protected]}
|
|
}
|
|
\date{December 2024}
|
|
|
|
\begin{document}
|
|
|
|
\hanzocoverpage
|
|
|
|
\begin{abstract}
|
|
We present \textbf{Hanzo ACI (AI Chain Infrastructure)}, a blockchain architecture purpose-built for decentralized AI compute markets. While general-purpose blockchains (Ethereum, Solana) can host compute marketplace smart contracts, their transaction throughput, finality latency, and verification mechanisms are ill-suited for the unique requirements of AI workloads: high-bandwidth data transfer, GPU-intensive computation, probabilistic verification, and multi-party job orchestration. ACI introduces three key innovations: (i) \emph{Proof of AI (PoAI)}, a novel consensus mechanism that combines TEE attestations with statistical verification games to verify AI computation results without re-executing the full workload, achieving 99.7\% detection rate of malicious providers at 0.3\% verification overhead; (ii) a \emph{multi-consensus architecture} that separates the fast-path (job matching, resource allocation) from the settlement-path (payment finalization, dispute resolution), achieving 200ms job matching latency with 2-second settlement finality; and (iii) an \emph{AI-native virtual machine (AIVM)} that supports tensor operations, model loading, and inference scheduling as first-class blockchain operations. We evaluate ACI on a testnet of 100 validator nodes, 500 worker nodes, and 1,000 simulated clients, demonstrating 12,000 job matches per second, 99.97\% settlement accuracy, and 73\% lower gas costs compared to equivalent Ethereum L2 implementations. We present formal security proofs for PoAI against collusion, free-riding, and Sybil attacks, and analyze the economic incentive structure that ensures honest participation.
|
|
\end{abstract}
|
|
|
|
\input{sections/01-introduction}
|
|
\input{sections/02-system-architecture}
|
|
\input{sections/03-proof-of-ai-poai}
|
|
\input{sections/04-multi-consensus-architecture}
|
|
\input{sections/05-ai-native-virtual-machine-aivm}
|
|
\input{sections/06-security-analysis}
|
|
\input{sections/07-token-economics}
|
|
\input{sections/08-evaluation}
|
|
\input{sections/09-related-work}
|
|
\input{sections/10-discussion}
|
|
\input{sections/99-conclusion}
|
|
|
|
\end{document}
|