Quick overview of Gensyn, a blockchain AI computing protocol led by a16z

Overview of Gensyn, a blockchain AI protocol led by a16z.

By Dong Nuo. On June 12, Gensyn, a blockchain AI computing protocol based in the UK, announced that it had completed a $43 million Series A financing round led by a16z. Gensyn has taken the lead in this AI revolution and has provided us with a response.

Gensyn is essentially a first-layer proof-of-stake blockchain based on the Substripe protocol. It can promote task allocation and rewards for machine learning through smart contracts, enabling rapid learning capabilities for AI models and lowering the cost of deep learning training. The cost of a single GPT-3 training session in 2020 was about $12 million, more than 270 times higher than the estimated $43,000 for GPT-2 training in 2019. Generally, the model complexity (size) of the best neural networks doubles every three months. The hourly cost of Gensyn’s machine learning training work is about $0.4, far lower than the required costs of AWS ($2) and GCP ($2.5). Gensyn wants to use blockchain and other technologies to achieve an efficient computing protocol for decentralized large-scale distributed deep learning, with probabilistic learning proofs and cryptocurrency incentive mechanisms.

Gensyn connects developers (anyone who can train machine learning models) with solvers (anyone who wants to use their own machines to train machine learning models). By utilizing idle computing devices with machine learning capabilities from around the world (such as small data centers, personal gaming computers, M1 and M2 Macs, and even smartphones) and connecting them to a global machine learning supercluster, the available computing power for machine learning is increased by 10-100 times. At the same time, Gensyn achieves trustless training of neural networks at a large scale and low cost through its innovative verification system and computing power supply.

1. Innovative verification system

The core challenge for Gensyn is to verify whether the computing tasks executed on the devices have been correctly executed and trigger payment through tokens. The Gensyn system mainly solves the verification problem through three concepts, including probabilistic proof-of-learning, graph-based precise positioning protocol, and Truebit-style incentive game.

Consists of four main participants, including submitters, solvers, validators, and whistleblowers. Submitters are the end users of the system, providing tasks to be calculated and paying for completed work units. Solvers are the main working part of the system, performing model training and generating proofs for validation. Validators link the non-deterministic training process to deterministic linear computation, replicate part of the solver’s proof, and compare distance with the expected threshold. Whistleblowers are the last line of defense, checking the work of validators and challenging for accumulated bonuses.

The system can accomplish all of these tasks without the need for trust and the cost is proportional to the size of the model, thus keeping validation costs constant. The innovation of the system lies in combining model training checkpoints with probability checks terminating on the chain, effectively solving the state dependence problem in training arbitrary scale neural networks.

2. New Computing Power Supply

The Gensyn system will use underutilized and unoptimized computing device resources. These devices include game GPUs that are currently unused and GPU mining machines from the pre-Ethereum PoW era. Because the protocol is decentralized, meaning that it will ultimately be managed by a majority of the community and cannot be “shutdown” without community consent, unlike web2, it is censorship-resistant. The innovation of the protocol lies in fully utilizing underutilized computing device resources, providing the community with more computing power, while also providing a new source of income for those who have unused devices. And the Gensyn protocol offers costs similar to GPUs owned by data centers, with a scale that can exceed AWS.

In summary, Gensyn’s core goal is to achieve AI democratization through decentralized planning, enabling more people to participate in the innovation and application of AI technology. The core idea of this plan is to utilize underutilized computing device resources, improve the efficiency and accuracy of AI models by building an open, decentralized validation system, and provide more opportunities and possibilities for AI entrepreneurs. It is an innovative and forward-looking plan that is expected to play an important role in the future AI field.