NuNet Catalyst Fund 10 Proposals
Greetings NuNetopians,
It has been no secret that NuNet’s development has shifted almost entirely to the Cardano blockchain, with our current testing occurring on the Cardano testnet with the aim to launch on Cardano mainnet later this year. This is due in no small part to the generous approval by the Cardano community of our Catalyst Fund 7 and 8 proposals.
NuNet’s Fund 8 proposal is the Decentralized GPU ML Cloud which is currently being tested, but all its deliverables have been developed. Fund 7’s Decentralized SPO Computing has been on hold, but NuNet will continue to work towards its completion, and we will provide an update about it in the near future.
As NuNet is nearing its mainnet launch, and due to the fact that Cardano’s Catalyst funding initiative is starting its tenth round after a considerable hiatus, NuNet has submitted a total of six proposals for the funding round. This article will give a brief overview of each of our proposals to introduce the use cases they will bring to NuNet’s decentralized cloud computing network as it continues to grow. The title of each section is also a link to the full proposal.
NuNet: Decentralized GPU Clusters — Research & PoC
As NuNet is currently capable of using single graphics processing units (GPUs) to run machine learning (ML) and other computational workflows, the next logical step is to research and develop a prototype to add the capability to utilize clusters of connected GPUs, such as GPU mining rigs along with single GPU systems. This use case will utilize three main components: containerization, resource allocation and peer-to-peer networking. For example, the Go Docker SDK will allow containerisation of jobs, Horovod would facilitate resource allocation, and Go libp2p would allow native peer-to-peer and decentralized network communication. If approved, this use case will allow NuNet to have a much higher potential level of computing power, as single compute providers are able to connect all their GPUs together on the network. In addition to NVIDIA, cross-vendor GPU support for AMD and Intel GPUs would play a significant role in scaling computation across different computing devices.
NuNet: Decentralized GPU Splitting On Software Level
This next proposal sounds similar to the previous, but it is very distinct. While GPU clusters would allow a single compute provider to connect their entire mining rig, GPU job splitting would allow single jobs to be replicated among multiple providers in the network, so that one device going down will not stop the entire process. This solution will deploy the prototype from the previous proposal mentioned above, into the current NuNet infrastructure. Once complete, it will allow service providers to run their computing jobs on many devices to help prevent their service from being interrupted unexpectedly.
NuNet: Running OSS ChatGPT Alternatives On Decentralized Network
Recent hype around AI has seen a consolidation around big players in terms of accessibility to models and computing power. To secure diversity we need to provide access to decentralized computing. NuNet will enable training, running and inference of various open-source chatbots on decentralized hardware. NuNet’s solution is all-inclusive for users and providers. For the development we shall use only open source frameworks as follows: OpenSource ChatGPT alternatives (e.g. Google Flan-T5 Large, Open Llama 7B, etc.)
NuNet: Bioinformatics Simulations With Decentralized Computing
There is growing demand for machine learning and data simulations in the biology and medical fields. We propose to add the capability for bioinformatics, medical, physics, and other research oriented simulations on NuNet. This use case will use open-source data science frameworks such as Anaconda to leverage scientific and other research tools, with CellModeller: a multicellular modeling framework as a real world example. This will enable running compute workflows on the network to support science and medical research such as those being conducted by Rejuve, a fellow SingularityNET ecosystem member, which focuses on longevity.
NuNet: DIDs For Components In NuNet Ecosystem
Despite the preference for anonymity that exists in the crypto industry, there are still times that it is useful and required to prove one’s identity. In NuNet’s case, for example, the General Data Protection Regulation (GDPR) law in the EU makes it so that EU citizens’ data cannot be stored outside of the EU, so for a compute provider who lives in the EU to legally serve other EU citizens, they would need to prove that they are located in the region, and the users would need to prove their citizenship.
NuNet’s “DIDs For Components In NuNet Ecosystem” Catalyst proposal aims to solve this complicated problem with decentralized identities (DIDs) by providing a mechanism for providers, consumers, services and hardware devices to uniquely identify themselves using DIDs and securely attach them to KYC (know your customer) and KYB (know your business) data and service providers. NuNet will accomplish this by using only open-source frameworks such as Aries Framework, W3C DID and IAMX for KYC service specifications and supporting documents. Furthermore, protecting the calls in the ecosystem is important, so to make the platform a safe place for our users we will be implementing an authorization layer using UCAN.
NuNet: Enabling Elasticsearch Clusters On Decentralized Hardware
A vital ingredient to machine learning is data, and while a great deal of data is available, it’s not always easy to find the appropriate data to train a specific algorithm. Elasticsearch is a fast and scalable search and analytics engine, providing powerful analytics and visualization tools; we propose that NuNet should enable running this on our decentralized network to allow access to large datasets. This will decrease reliance on big tech databases, and enable anyone with eligible devices to host Elasticsearch nodes, configure them into clusters and be rewarded via Cardano smart contracts. Open-source frameworks to be used for this include Docker, Elastic Stack and Go libp2p.
ML-Powered NFT Recommendation API, Trained on the NuNet Network
ArgusNFT is a project on Cardano which seeks to protect against intellectual property theft in non-fungible tokens (NFTs) by using machine learning to distinguish between original, similar and counterfeit tokens. They propose to open-source code pertaining to the training pipeline of their ML model, which will run on NuNet. This solution will be an NFT recommendation API to be used by NFT marketplaces for search and discovery. Their solution is unique because it combines the power of public wallet data, NuNet’s decentralized computing network, and their extensive experience implementing ML solutions in ecommerce.
NuNet Is Hiring!
NuNet currently has a number of open positions for various roles within the team. If you have the skills and desire to join us in our journey, you can find more information and contact us through our career page.
About NuNet
NuNet lets anyone share and monetize their compute resources, turning cloud computing power from a centralized service into an open protocol powered by blockchain. Find out more via:
