Universal Basic Compute for Actually Open AI
summary
Lilypad is a live, permissionless distributed compute network with a Milky Way testnet, EVM compatibility and an ERC‑20 gas token, claiming to run internet‑scale AI/ML jobs via a marketplace of deterministic modules. Jobs are module = container + spec (Docker/WASM) so outputs are reproducible; verification uses optimistic reproducibility with collateral staking. Provided modules already run Stable Diffusion, text→video, text→speech, Llama 3 and AlphaFold2; CLI, new JavaScript wrapper and a Hardhat smart‑contract template are working for automated job submission.
AI-written summary · about
chapters (13)
- 0:00Introduction to Universal Basic Compute
- 1:23Open infrastructure for AI
- 3:52LilyPad network goals
- 5:38How LilyPad works
- 7:28Running compute jobs
- 12:10Live demo of LilyPad
- 17:30JavaScript wrapper for LilyPad
- 19:14Contribute compute power
- 20:39Custom LilyPad modules
- 23:09Hackathon project ideas
- 24:54Hackathon opportunities
- 26:59LilyPad next steps
- 27:42Q&A
related talks
37mUniversal AI: Execute Your Models Where Your Data (And Users) Are
38mIntroducing Open Platform for Enterprise AI
47mOwn Your LLM: Running Generative and Agentic AI Locally
15mBuilding AI Agents the Web3 Way
30mAI Application Architecture: Building for Scale and Efficiency
19mBAML: a new programming language for AI
01h 42mBuilding Responsible AI Agents in Decentralized Open-Source Environments
43mLLMs in the Wild - Streaming, RAG, and Real Time GenAI at Scale