Decentralized Inference

Nesa's decentralized inference process is the cornerstone of our autonomous AI oracle network, enabling the first trustless environment where AI computations are performed transparently and reliably on-chain.

This section outlines the design of Nesa's decentralized inference framework, which is composed of several core components: users who submit inference requests, chain contracts responsible for the verification and aggregation of results, and nodes that process these requests.

This framework leverages a two-phase transaction structure, utilizing a commit-reveal paradigm, to safeguard against dishonest behavior and free-riding. This ensures that nodes are incentivized to perform their computations honestly and that users can trust the integrity of the inference results.

The system maintains a decentralized approach by employing smart contracts for the key processes of verification and aggregation, allowing for a scalable network that harnesses the collective computational power of its participants.

Model Partitioning and Deep Network ShardingCache Optimization to Enhance EfficiencyBSNS with Parameter-efficient Fine-tuning via AdaptersDynamic Sharding of Arbitrary Neural NetworksEnhanced MTPP Slicing of Topological OrderSwarm Topology