gonka-ai/gonka

What it solves

It addresses the high cost and monopolistic nature of centralized cloud providers by providing a decentralized infrastructure for AI model training and inference. It aims to optimize computational power so that resources are used for actual AI workloads rather than being wasted on blockchain security.

How it works

The system uses a novel consensus mechanism called Proof of Work 2.0, where nodes compete in "Sprints" using transformer-based models to solve AI tasks. A node's success in these races determines its voting weight, which in turn controls how many AI inference and training workloads it is assigned.

To ensure honesty without excessive overhead, the platform employs randomized task verification (checking only 1-10% of tasks) and a reputation system that reduces verification frequency for consistently honest nodes. For distributed training, it leverages DiLoCo's periodic synchronization to minimize communication overhead across geo-distributed hardware providers.

Who it’s for

• Developers who want to run LLM training and inference on a decentralized network.
• Hosts (Hardware Providers) who wish to contribute their computational resources to earn rewards.

Highlights

• Proof of Work 2.0: Aligns blockchain consensus with actual AI model workloads.
• Randomized Verification: Reduces system overhead by probabilistically verifying a subset of tasks.
• Geo-Distributed Training: Uses DiLoCo to enable scalable, decentralized training with minimal communication lag.
• Reputation-Based Trust: Dynamically adjusts verification frequency based on a node's history of honest behavior.