BasedAI (BASEDAI)
BasedAI is a decentralized peer-to-peer network designed for Zero-Knowledge Large Language Models (ZK-LLMs) and smart contracts, founded by Sean Wellington. It integrates Fully Homomorphic Encryption (FHE) with large language models (LLMs), facilitating secure and private AI computation without compromising data privacy. The network utilizes a distributed system known as "Cerberus Squeezing" to transform traditional LLMs into encrypted zero-knowledge LLMs (ZK-LLMs). This method optimizes both performance and privacy by enabling miners to process user interactions without decrypting inputs or outputs. The platform also combines diverse blockchain technologies, drawing on elements from Polkadot’s Substrate, Ethereum’s EVM, and GPU configurations from RNDR, to support smart contracts and asset management.
| Ticker | BASEDAI |
| Category | Artificial Intelligence (AI) |
| Website | https://getbased.ai |
| @getbasedai | |
| Contract Addresses | |
|---|---|
| ethereum | 0x44...85 Copied! |
Technology
BasedAI employs Fully Homomorphic Encryption (FHE) to ensure data remains encrypted during storage, transmission, and computation, allowing computations on encrypted data without decryption. Central to its ecosystem are processing nodes known as "Brains," which execute tasks associated with LLMs and form the decentralized infrastructure for privacy-focused AI processing.
Testnets and development
BasedAI's development includes two main testnets: Prometheus and Cyan. The Prometheus testnet has been completed, and the Cyan testnet, launched in June 2024, functions as a release candidate to test the integration of Brain NFTs and related tokens. This phase focuses on token emissions, rewards, and administration, leading up to the mainnet launch.
BasedAI ecosystem
BasedAI operates on a decentralized blockchain network, emphasizing data privacy and security in AI applications. It integrates zk-LLMs with homomorphic encryption to provide a secure AI environment and decentralize AI model ownership. Users interact with the network via the Based Command Line Interface (Based CLI), which facilitates functions like wallet management and interaction with computational nodes.
Key features
- Cerberus Squeezing: Converts traditional LLMs into zk-LLMs for secure, encrypted AI processing.
- Fully Homomorphic Encryption (FHE): Keeps data encrypted throughout its lifecycle while supporting computations.
- Based CLI: A command line interface for interacting with the platform and managing wallets.
- Testnet Phases: Includes Prometheus and Cyan for testing functionality before full deployment.
Applications
BasedAI enhances privacy in AI processing, democratizes AI model ownership through Brain NFTs, and provides a secure environment for deploying smart contracts. Its integration of AI and blockchain technologies supports decentralized AI-powered tools, secure data processing, and privacy-preserving computations.
Tokenomics
Rewarding participation: Incentives and token emissions
BasedAI incentivizes ecosystem participation with $BASED tokens. Every 10 seconds, 10 $BASED tokens are distributed among active Brains, rewarding computational tasks based on staked $BASED and validator performance. The top 30% of Brains earn additional bonuses, fostering competition and efficiency. All $BASED tokens can be bridged with the wrapped BasedAI ERC-20 token on Ethereum.
Emission halving schedule
To manage supply, BasedAI implements an annual halving schedule that reduces block rewards. Starting in April 2024, rewards will halve from 10 $BASED tokens to 5, with subsequent halvings each year, balancing supply and maintaining incentives.
Governance
Staking and validator incentives
BasedAI's staking mechanism allows users to stake $BASED tokens to any Brain or validator. Validators maintain network functionality by processing computational tasks, and stakeholders earn rewards based on emissions allocated to their chosen validators.
Allocation of rewards
Seventy-five percent of emissions go to validators and miners performing essential tasks, while 25% is allocated to Brain owners or specific validators managing the Brain.
Owner adjustments and stake-weighted rewards
Brain owners can adjust reward distribution within nodes, creating a competitive environment where validators are rewarded based on performance. Stake-weighted incentives provide larger rewards to participants with significant investments in specific Brains.
Utility
Brain NFTs and validator diversification
BasedAI encourages validator diversification by rewarding the top 70% of validators, promoting stake distribution across multiple Brains and reducing centralization risks. Validators cannot duplicate stakes across Brains, supporting independent Brain operations.
Mitigation of centralization
To decentralize the network, BasedAI caps the maximum stake per Brain, preventing any entity from controlling more than 0.5% of the network's total stake in $BASED, ensuring equitable reward distribution.
Temporal Fusion Transformer (TFT) Enforcer
The Temporal Fusion Transformer (TFT) Enforcer optimizes reward distribution by analyzing network data over time. It evaluates factors like transaction volumes and block times, dynamically adjusting incentives to validators and Brains for fair reward distribution.
Sean Wellington
Sean Wellington, a researcher at Based Labs with an educational background from UC Berkeley, contributed significantly to BasedAI's development. He focused on integrating Fully Homomorphic Encryption (FHE) with large language models (LLMs) to maintain data privacy without decryption. His work includes developing the "Cerberus Squeezing" mechanism to optimize performance in FHE-compliant environments. Wellington's research spans cryptography, decentralized systems, and AI, with numerous patents and publications highlighting his contributions to distributed data management and privacy-focused AI technologies.