Hyper Oracle Introduces opML to Run Large ML Models on Ethereum Blockchain

53 views 12:16 am 0 Comments November 22, 2023

 

Quick Take

Hyper Oracle’s Optimistic Machine Learning (opML) launch marks a significant stride in blockchain technology, offering the first open-source solution for efficiently running large ML models on Ethereum. This breakthrough combines flexibility, performance, and cost-effectiveness, propelling AI and ML integration into new realms of blockchain applications.

Hyper Oracle’s Optimistic Machine Learning

Hyper Oracle has launched its Optimistic Machine Learning (opML) technology, marking the first open-source implementation of this advanced solution. This initiative is set to revolutionize how considerable machine learning (ML) models are run on the Ethereum blockchain, bringing unprecedented flexibility and performance.

Hyper Oracle

Hyper Oracle is known for its innovative approach to programmable zero-knowledge oracles. It has been at the forefront of integrating zero-knowledge proofs (ZKP) with blockchain security and decentralization, offering solutions ranging from indexing to intelligent contract automation​​​​.

Earlier this year, Hyper Oracle received a significant boost with a $3 million pre-seed funding round co-led by Sequoia China and Dao5. This funding was instrumental in advancing the firm’s research and development in ZK Oracle and blockchain infrastructures​​.

Opel: A Leap Forward in Onchain AI and ML

The introduction of opML by Hyper Oracle is a significant milestone in the blockchain realm. This technology stands apart with its optimistic verification mechanism, which offers enhanced performance and flexibility compared to zkML (zero-knowledge Machine Learning).

opML’s capability to run large ML models, including those as complex as GPT 3.5, on the mainnet is a testament to its advanced engineering. Its low-cost, high-efficiency model allows it to operate large language models on basic hardware like laptops without requiring specialized resources such as GPUs.

The Importance of opML in the Blockchain Ecosystem

Hyper Oracle’s opML supports the inference of ML models and extends its versatility to fine-tuning and training processes. This broad applicability makes it a powerful tool for various ML tasks. The technology’s integration into the blockchain offers several benefits, including enhanced fairness, transparency, decentralization, and the assurance of on-chain validity for ML models.

Comparing HTML and XML

While both opML and zkML aim to verify the validity of computations in ML models, they differ significantly in their approach and capabilities. zkML relies on zk proofs, which, despite offering high security through mathematics and cryptography, face limitations in terms of performance. It struggles with constraints like memory usage, quantization, and circuit size limit, rendering it suitable only for smaller models. In contrast, opML’s optimistic approach overcomes these limitations, making it feasible to implement larger models effectively.

The Road Ahead for Hyper Oracle and Blockchain Technology

With this launch, Hyper Oracle has positioned itself as a pioneer in bridging the gap between advanced AI/ML and blockchain technology. The advent of opML signals a new era in on-chain artificial intelligence (AI) and machine learning, introducing increased scalability, efficiency, and decentralization. These advancements aim to maintain high standards of transparency and security in the field.

Hyper Oracle’s journey, characterized by innovation and strategic funding, showcases its commitment to enhancing blockchain technology’s scalability and utility. In the dynamic evolution of the blockchain industry, the contributions of Hyper Oracle, especially with its recent opML technology, are poised to play an essential role in shaping the future landscape of integrating blockchain and artificial intelligence/machine learning (AI/ML).

This article was generated with the support of AI and reviewed by an editor. If you would like more information, you can see our T&C.