Reasonable enough, will try my best thank you.
3 Likes
Thanks a bunch. Much appreciated
This is cool! Just to firm my understanding of this…
One could become a node provider by running a node locally & ultimately get paid for running this node IF other folks use this node for it’s compute.
Obviously used servers(ETH miners) would make the most sense; because the cost has already been amortized and because of global supply of GPU issues. Also some clever LLMs can make use of CPU+GPU; so that again makes sense.
All book keeping, proof of providing useful compute etc from EMC as layer 2 would be settling in IC as layer 1.
Did i get that right?
3 Likes
Thanks for the feedback. Indeed it is impressive.
Btw my first camera was a b/w Agfa. Brought back so many memories!
1 Like
To be fair and consistent this request for moderating interpersonal comments should apply to @zed.emc too.
Way back in this thread, before the discussion degraded, I found this this explanation of the very different environment the EMC Dev team are working under:
@zed.emc could you explain a little more about how your technical team works given these constraints (unfamiliar to many of us). For example, does EMC team keep shared code repositories on your own servers because of this legal environment rather than “developing in the open”?
Are there impacts on sharing code on public internet platforms such as GitHub due to PRC export controls such as cryptographic security technology, artificial intelligence interactive interface technology referred to the last row of the table in section 1 of that linked article?
If this public forum isn’t the best place to share specific answers to those questions then maybe DM to @Icdev2dev might assist and more generic statements might be shared here.
For those interested, it is worth reading that whole article. Also the “About” section of that site might be of use to anyone assisting behind the scenes.
1 Like
Just to be clear I have nothing against any of these arguments, yes it can be tough for them for sure. But at the same time I am personally working with people from Alibaba wholesale for 10 years now and not only do they speak and write decent English but are able to use apps like Skype, Whatsapp and so forth. Let’s say they are allowed since they work in sales and export. I now have with me here in Budapest someone from Alibaba Cloud who teaches at ELTE Faculty of Informatics and will drop in for the AMA after the EMC team finishes in their time, no rush. Best of luck take your time we the community all want you to succeed.
2 Likes
Sounds like an excellent idea
1 Like
Will ask in Standard/Simplified/Traditional Chinese by a native in live recording and will translate it after so it will be easy for them.
2 Likes
Project Name: LingChuang & MetaWhale
Founders: Xender & Mowen, pioneers combining NFT and AIGC
Product Introduction: LingChuang AI, a Wechat mini-program, currently has over 200,000 users and a community of 20,000+. The web3 project MetaWhale has initiated its collaboration with AngryCat for the second phase of AIGC derivative NFT distribution. The LingChuang AI platform for PC is currently under development.
Project Name: AI Metafun
Founders: Barry & Shadow, pioneers combining AIGC and e-commerce
Product Introduction: Currently collaborating with Hashmeta (based in Singapore), providing AI-generated content tools and AI models/applications for verticals such as beauty, maternity, and home life to over 7 million e-commerce platform KOLs. Currently under development.
Project Name: Earth Open-source Content Generation Platform
Founder: Shadow, Founder of Mixlab, AIGC Development Alliance
Product Introduction: Earth is an open-source content generation platform that has integrated e-commerce applications such as MetaFun and Audio AI lab. Initial products have entered testing, focusing on audio applications such as music and audiobooks.
Unbounded Community - Mixlab
Mixlab is an interdisciplinary organization consisting of over 70,000 designers and engineers. As a next-generation innovation organization, we have designed an incubation mechanism to ensure the integration of inspiration and knowledge from different disciplines, helping members materialize their ideas and drive value innovation. As a content community, we welcome talents from diverse disciplines and advocate for cross-disciplinary collaboration, innovation, open-source, experimentation, and decentralization, providing a conducive environment for startups and individual innovators to kickstart their projects.
We have just open-sourced some AI demos. Prior to the launch of SNS, all voting and governance mechanisms will continue to be iterated and improved, and more codes will gradually be made open-source.
In terms of AI applications, more and more partners are beginning to integrate, including Metaverse, Crypto Analysis, and others. We are also actively preparing for the AI+Web3 Global Hackathon in July.
As to AI computing power, EMC has already partnered with several Filecoin miners (sealing machines), and there are several Ethereum miners currently in the process of integration, with thousands of idle 30-series GPUs in their possession.
6 Likes
Haha, we also have engineers from Alibaba on our team. Bitcoin and Ethereum have shown the world that we can transcend national boundaries and collaborate together. Although part of the EMC team is based in China, our work is not limited in any way. The EMC Foundation is in Singapore, and it is inherently global. We hope to leverage our capabilities to integrate Web3 and AI, and expand the markets for ICP and EMC in Asia.
2 Likes
So according to today’s global R&D slides (May 31, 2023) EMC SNS is coming in June. EMC team and @zire is that correct?
1 Like
My answer is YES.
In short, the onchain AI approach in our case involves two main steps. Firstly, we design the model and conduct experiments to test the effectiveness of the AI. This initial phase is carried out on our own instance. Secondly, we deploy the AI for production in either the AWS or EMC environment.
Typically, ML developers spend the majority of their time on step 1, which is focused on research and development. Step 2, on the other hand, is more akin to engineering work, involving the deployment of the AI in a production environment.
Currently, we are intensively focused on step 1 as we develop our next vision AI. Additionally, I am considering trying out the EMC network by deploying our previous model to explore the possibilities of combining AI with Web3.0.
3 Likes
Thank you much for your feedback!
A couple of quick questions:
since high fidelity music is pretty intense from a load perspective (even at inference time), what are the typical compute power requirements do you see (GPU standpoint)?
What are the differentiating factors that would lead you to consider EMC vs say runpod?
EMC Opensource at EMCProtocol-dev · GitHub
Includes node source code, SDK, test cases, etc. Today we will be testing the IC contract reward distribution. If it passes, we will open source the IC contract portion tomorrow.
8 Likes
Hi IC community,
Here is the EMC Github source code: EMCProtocol-dev · GitHub. We welcome more developers to follow, audit, and contribute to the codes. We will launch a bounty board around EMC to attract global developers’ participation. Interested developers can join the TECHNOLOGY COMMUNITY on Discord（Discord | Your Place to Talk and Hang Out and mention @Flash.EMC
EMC Protocol is a blockchain and peer-to-peer network-based edge computing protocol that includes features such as smart contracts, decentralized storage, and consensus mechanisms for transaction verification.
emc_node ( GitHub - EMCProtocol-dev/emc_node**)**
The EMC node is the core protocol of the EMC computing power network. It utilizes the EMC contract based on IC to interact with nodes across the network (validation nodes, intelligent routing nodes, computing power nodes). Using a state machine, it performs task accounting for the EMC Layer, enabling computational scheduling, application deployment, task accounting, and rewards across the EMC network. For more technical details, please refer to: https://edgematrix.pro/technology.
Openverse Framework is a web3 dapp development framework built on top of the EMC protocol. It provides a rich set of SDKs and APIs to help developers develop, deploy, and manage applications more easily.
EMC-Requester（ GitHub - EMCProtocol-dev/EMC-Requester ）
Deployment testing for computing power nodes and development testing for AI computing applications.
emc_java_sdk（ GitHub - EMCProtocol-dev/emc_java_sdk ）
edgematrixjs-monorepo-sdk（ GitHub - EMCProtocol-dev/edgematrixjs-monorepo ）
The above are Java and JS SDKs that developers can use to interact with EMC’s computing power network. By selecting suitable computing power nodes, developers can deploy and run their own models and applications. We will gradually release SDK versions in multiple languages, such as Python, Go, Rust, and Motoko. We welcome more developers to contribute and join us.
EMC-SD（ GitHub - EMCProtocol-dev/EMC-SD ）
This AI demo is based on the EMC computing power protocol and is a use case of stable diffusion and its interface is deployed in an IC canister. Developers can use this demo to gain a better understanding of the principles, applications, and scalability of EMC. Currently, there are mature online AI drawing programs running on EMC using this approach, achieving a cost that is 1/10th of the official pricing of cloud services and realizing new green energy solutions.
The contract source code for EMC network rewards, transactions, and other tokenomic features based on IC (EMC tokenomic) is currently being deployed and audited. It will be gradually open-sourced before the SNS launch.
11 Likes
Thank you for starting the process of open sourcing! I had a look at emc_node and it seems that it is built on GitHub - 0xPolygon/polygon-edge: A Framework for Building Ethereum-compatible Blockchain Networks. So EMC nodes still employ their own consensus protocol (based on GitHub - 0xPolygon/polygon-edge: A Framework for Building Ethereum-compatible Blockchain Networks) ? Looking forward to having a look at the IC parts or to reading a deeper dive on how this integration works.
9 Likes
Currently, our inference approach involves renting a computational instance on AWS 24/7 (without mentioning the cost of training). However, this approach is very expensive, as I attached by the following costs.
This is not a scalable option for startups like us who don’t have access to millions of dollars in computation credits like OpenAI.
In my pursuit of a more cost-effective solution, I have spent years exploring and searching for alternatives. I came across some P2P services like Mist and Vast, but I have concerns about their stability and security compared to AWS. It would be disastrous for us to deploy our product on a random GPU server only to have the service provider disappear or steal our source code.
Like my previous post, I have always wondered if we could leverage abandoned Ethereum PoS GPUs to train AI in an affordable, secure, and stable manner. This is where blockchain and EMC might come into play.
Furthermore, if EMC can offer seamless AI OS management similar to ICP for dapp as an L2 solution, it would be an added benefit for us. I won’t delve too deeply into this aspect at the moment.
2 Likes
Have stated the same and still nothing changed … the usual " probably nothing" applies.
So yeah, just give them more time and spread the love 
The definition of Stable Coin must be confusing since ckBTC is not one.
3 Likes