As AI advances, the concept of "everyone is a scientist" becomes attainable. Vertical AI is expanding human capabilities to scientist level across multiple disciplines. However, centralized control of AI resources and insufficient open-source incentives hinder the collaboration, and equitable value distributions.

While open-source AI for science promotes collaboration, contributors face fragmented access and lack equitable rewards. Existing blockchain solutions often underestimate the complexities of co-building AI, relying on centralized servers, closed point systems, and governance models influenced by major token holders. Consequently, off-chain AI contributions—such as code commits on GitHub or user feedback—remain under-incentivized and poorly recognized.

Hetu offers an AI4Science OS built on decentralized model graphs, coupled with causal dependency tracking. This enables impartial incentives for vertical model creators and contributors:

• Decentralized Model Graphs: Transparently link and verify AI models and data, ensuring traceable contributions with 1 million QPS/subgraph.
• Causal Dependency Tracking (POCW): The Proof of Causal Work consensus accurately evaluates each contributor’s work, enabling fairer incentives with 100x throughput.
• Hyper-Scalable Consensus: Hetu’s DAG-based layer 2 achieves over 160,000 TPS with sub-second confirmation—suitable for large-scale scientific models.

Hetu's architecture enables verifiable and transparent auditing of open-source model contributions, rewarding teams and individuals in a purely decentralized manner. This fosters broader collaboration and continuous innovation.

Hetu bootstraps open-source vertical models for science in tokenization model and service model:

1. 🚀For fair and cost-effective attributions of ModelDAOs for science, Hetu provides instant accountability through verifiable model graph dependencies.
2. ⚡️For broader and fairer incentives for ModelDAOs of science, Hetu provides verifiable impartiality via its hyper-scale modelchains.
3. 🛡️For impartial and adaptive governance of ModelDAOs for science, Hetu provides anti-censorship voting and personalized privacy.

• 🔗 Akasha
• 🔗 Prof. Li Jialin

Hetu is developed by Advaita Labs, with 10+ OSDI, SOSP, and NSDI publications in Causality and AI systems.

Guided by the philosophy of intersubjective consensus, Hetu aims to drive breakthroughs in the deep integration of humans and AIs.

Hetu Key Research of VLC:

• Chrono: A Peer-to-Peer Network with Verifiable Causality
• Building a Verifiable Logical Clock for P2P Networks

Hetu Key Consensus Research:

• 📄 NeoBFT: Accelerating Byzantine Fault Tolerance Using Authenticated In-Network Ordering (OSDI'20)
• 📄 Just Say NO to Paxos Overhead: Replacing Consensus with Network Ordering (OSDI'16)
• 📄 Pegasus: Tolerating Skewed Workloads in Distributed Storage with In-Network Coherence Directories (OSDI'20)
• Proof of Causal Work (POCW)
• Intersubjective Vision Paper

Hetu Key AI Research:

• https://yao.lu/ai_native.pdf
• https://dl.acm.org/doi/10.1145/3698038.3698563
• https://proceedings.neurips.cc/paper_files/paper/2024/hash/9482f45fdd89aba9130bb04c44f788a9-Abstract-Conference.html

Hetu Lightpaper

Hetu Deck

Hetu Website