OpenGradient Architecture

OpenGradient's novel Hybrid AI Compute Architecture (HACA) leverages node specialization to create a powerful, versatile platform.

NOTE

HACA allows OpenGradient to provide efficient and scalable inference capabilities while remaining fully verifiable and decentralized

HACA

HACA is made up of the following core components:

• Full Nodes: Blockchain full nodes that execute EVM transactions for the blockchain. This handles native precompiles for and computing the full state-transition function (STF).
• Inference Nodes: Stateless nodes with specialized hardware designed and optimized to run inference and generate any associated zkML proofs or TEE attestations. These nodes can have a variety of hardware architectures and can consist of CPUs, GPUs, TPUs, or software-based TEEs.
• Storage Nodes: Hard disk storage nodes are designed for our decentralized filestore, which stores cold models uploaded onto the network.
• Data Nodes: Enclave nodes that leverage the TLS protocol through Nitriding. TLS backed encrypted connections are established between a Data Node and third-party services, and the Data Node can securely access arbitrary data sources and create data access services that smart contracts on OpenGradient can access. Developers can push to/pull data from any external resources directly from smart contracts.

Below is a diagram describing the HACA components and workflow in the OpenGradient network.