Hetu is the intersubjective consensus, decentralizing AI for Science. It's the decentralized model graph with causal dependency, bringing impartial value distribution for accelerating science. Advaita Labs created it with 10+ OSDI, SOSP, and NSDI papers in causality and AI systems.

The value distribution of open-source AIs for Science is dispersed, isolated and inequitable. The current blockchain solutions aggravated monopoly and canalization during this process, which excludes common contributors from participating through GitHub or social networks.

Hetu implements decentralized modelgraphs with causal dependency graphs to generate impartial incentives for vertical model contributors. Its features include hyper-scalable verifications, interoperable sequencing incentivization, and anti-censorship governance.

• Hetu‘s modelgraph is enabled by causality graphs, by verifiable logic clocks with TEE.
• Hetu’s root consensus is intersubjective: proof of causality work (POCW).
• Hetu provides instant finality by hyperscalable layer 2 using DAG, with a theoretical TPS of 160,000+.

Hetu facilitates fair and impartial value distribution between off-chain AI models, enabling transparent and interoperable auditing for open-source science model contributors.

Hetu bootstraps the open-source vertical models for science cost-effectively and autonomously:

1. 🚀For fair attributions of ModelDAOs of science, Hetu provides instant accountability by verifiable dependency of the modelgraphs.
2. ⚡️For wider and fair incentives of ModelDAOs of science, Hetu provides verifiable impartiality via its hyperscale AI-native chains.
3. 🛡️For impartial governance of ModelDAOs of science, Hetu provides anti-censorship voting via its off-chain governance frameworks.

• 🔗 Akasha
• 🔗 Prof. Li Jialin

With the unique philosophy of intersubjective consensus, Hetu is envisioned to drive breakthroughs of deep entanglement between humans and AIs. Hetu unifies the power of decentralized science and community-first, to bring science assets to Ethereum.Cofounded in 2021, we have completed core research, testnet 1.0 and plan to release mainnet phase 1 in Q2 of 2025.

• Intersubjective Vision Paper
• Hetu Key Research of VLC
• Proof of Causality Work (POCW)

• 📄 Vault: Decentralized Storage Made Durable
• 📄 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)
• 📄 NeoBFT: Accelerating Byzantine Fault Tolerance Using Authenticated In-Network Ordering (OSDI'20)