The network construction

In decentralized service networks, blockchains fulfill two roles: they serve both as registers of cryptocurrency ownership and as foundations for decentralized services. In our system, the registration of participants and distribution of rewards happen on-chain, whereas the actual execution of services and model generations occur off-chain due to the inherent costs and limitations of on-chain operations. On-chain operations are not only slow and expensive, but also restricted, unable to benefit from real-world data and various functionalities that simply can't be accomplished on-chain. These functionalities include diverse forms of computation, fast data distribution between miners and clients, and flexible infrastructure upgrades, among other features.

To effectively leverage the potential of this decentralized network for AI training, a two-layer architecture is implemented: the on-chain component SC, which records the value flow in the network, and the off-chain component exec, consisting of a set of protocols running on the MintAI network (MAN) where utilities are performed. By securely integrating the on-chain functionality with the vast array of off-chain services offered by the MAN, it can exhibit the robustness and upgradability that traditional Layer 1 solutions often lack. In the L1-L2 design, the protocols and infrastructures primarily operate off-chain in the decentralized network, whereas token utilities such as transfer and withdrawal operate on Layer 2 of any mainstream blockchain. This setup allows the system to continuously update with additional features and utilities, while keeping the network's assets and user experience unaffected.