Q2 2020: The Raven Distribution Framework solves the deep learning problem that MIT researchers are warning the AI industry of
Dear Raven Protocol Community,
A few weeks ago, we launched our open source / public beta release of the Raven Distribution Framework. That was a huge milestone that sets us up for tackling the seemingly impossible problems that lie ahead in the progression of the Artificial Intelligence and Machine Learning industry.
In the past decade, we saw the great rise (some even call it a renaissance of sorts) of the AI industry due to advancements in hardware. It resulted in applying AI/ML techniques to almost every field and we saw the mega-corps dominate it.
However, MIT researchers warn that deep learning is approaching computational limits.
Progress in deep learning has been “strongly reliant” on increases in compute. It’s their assertion that continued progress will require “dramatically” more computationally efficient deep learning methods, either through changes to existing techniques or via new as-yet-undiscovered methods.
— Venture Beat’s The Machine: Making Sense of AI Series
Raven solves the problem of computational limits with our focus on the speed of AI training via the Raven Distribution Framework
Raven is one of a handful of organizations in the world working on changes in techniques, new approaches, and undiscovered methods that will drive the next AI Renaissance.
Our approach has no dependency on hardware and anyone can run a node from their browser to contribute AI compute to the Raven network. A decentralized and distributed deep learning training protocol means that these superpowers will be in the hands of the community and out of centralized powers.
The MIT researchers estimate that three years of algorithmic improvement is equivalent to a 10 times increase in computing power.
It is clear that we are in the right place at the right time. We have a shot at something that has never been done before. We’re still super early in building out the entire ecosystem, but you can see how fundamental Raven Protocol is and why it needs to exist. The AI/ML industry is reaching its limits and we’re unlocking new approaches and methods to push the industry into its next growth phase.
OpenAI recently released GPT-3 their text generating neural network that sounds like a human. That is an incredibly difficult problem and requires tons of compute resources. In fact, Raven started because our co-founder Sherman was trying to generate social media content and with his limited resources kept burning cash every month. He asked co founder Rahul how to solve that problem. One thing led to another and Raven was born.
However, the computational resources needed to actually use GPT-3 in the real world make it extremely impractical. So while the work is certainly interesting and insightful, I wouldn’t call it a major step forward for the field.
This is exactly the problem the raven distribution framework solves. We don’t have a dependency on hardware. You can contribute compute power on your browser.
Raven unlocking this value for the entire industry would be major step forward in the the AI field.
Checking out the Raven Distribution Framework
Goals of the Public Beta
- Allow users to run a local instance of the Raven Server and Raven Client that can communicate with each other.
- Test how simple computations are stored, passed, and processed between the Server to the Client.
- Provide the foundation to allow developers to plug in their own libraries or algorithms to train using the Raven Distribution Framework.
- Find bugs and security issues in the code base.
Do feel free to dive in, clone the repository, and make pull requests. Have a read through the Building Blocks of Raven blog post. You will be able to set up a Raven Server and a Raven Client using Docker.
After that, you can launch a demo app that will distribute computations like Matrix Multiplication (most operations while training a neural network require some form of MM) to the Raven Client, which sends the result back to the Raven server, and accessible by the demo app.
Voilà! We have the basis for decentralized and distributed deep-learning training with absolutely no dependency on hardware.
The Raven Community is getting super excited!
A lot more exciting things happened in Q2
You can only imagine what is in store for the Q3 and beyond. We’re really eager and dying to share it with you. We just can’t publicly announce any of it yet.
The Future of Raven Protocol
Raven can only exist due to the worlds of AI and Blockchain colliding. With the launch of the open source and public beta, we are seeing how important our infrastructure is for both the decentralized world and the AI community.
The more algorithms that are implemented and released on top of the Raven Distribution Framework.. means the more value we provide to real companies and users.
Exciting things on the Q3 Development Roadmap:
- Logistic Regression built on the Raven Distribution Framework. This will also serve as a blueprint for best practices on building algorithms on top of our framework.
- Linear Regression built on the framework.
- A demo web app for everyone to test out framework, to train Logistic Regression, Linear Regression, and to play with the Raven Distribution Framework. We will provide sample datasets.
- Beta customers on a high touch basis. Many B2B companies have a “professional services” arm. In addition to a path to revenue, this gives us validation and focus on what gets built and released on top of the Raven Distribution Framework. Big thanks to beta customers collaborating with us. We will ensure your business is a success too. If you are an AI/ML company interested in a framework that tackles the computational limits of AI training (our focus is on speed), then we’d love to hear from you!
- Start building out the network with our 18K+ RAVEN holders. Time to engage them to contribute compute power to the network. All you need is a browser. Get your RAVEN ready 🥩
- Start exploring governance of the network. Governance is especially important with respect to AI ethics. If GPT-3 were dangerous for the world and someone implemented it on Raven Protocol, RAVEN holders might want to have some say on whether or not to release it publicly on the network.
If you squint a little, the rewards + governance are starting to look a little like defi :) In all seriousness, many pieces of the puzzle are starting to align and come together. A lot of big things to come ahead. We know the community is eager to support us as node contributors, as liquidity providers, to push what we’re building to larger communities, and of course to use Raven Protocol itself. Let’s do this together!
Raven Protocol: Q2 2019 Tech and Community Update:
https://medium.com/ravenprotocol/q2-2019-tech-and-community-update-4f836a9a1e97
Raven Protocol: Q3 2019 Tech project development Update:
https://medium.com/ravenprotocol/tldr-raven-stayed-heads-down-building-in-q3-2019-ae5f242dc15d
Raven Protocol: Q4 2019 Tech project development Update:
https://medium.com/ravenprotocol/happy-2020-an-important-update-from-the-raven-team-67b5e88e1b1d
Raven Protocol: Q1 2020 Tech project development Update:
https://medium.com/ravenprotocol/q1-2020-success-during-hard-times-3dcdbb0faba8
Raven Protocol Project Review:
https://cryptocalibur.com/portfolio-item/raven-protocol-review
Raven Protocol White Paper:
https://drive.google.com/file/d/1FAaVKkg_CjxMj-n1yHZc6ufcVDtOU1Ct/view?usp=sharing
OFFICIAL CHANNELS:
Official Email Address: founders@ravenprotocol.com
Official Website Link: http://www.RavenProtocol.com
Official Announcement Channel: https://t.me/raven_announcements
Official Telegram Group: https://t.me/ravenprotocol
Official Twitter: https://twitter.com/raven_protocol
Official Medium: https://medium.com/ravenprotocol
Official LinkedIn: https://linkedin.com/company/ravenprotocol
Official Github: https://www.github.com/ravenprotocol
