Source: Rahul Vishwakarma & Sherman Lee, Raven Protocol Whitepaper (May 2019) avail. online.
Another team addressing the compute problem in AI/ML (see also Field Note #63 on Genysn)...we need more.
One-Liner: "Raven Protocol is a decentralized and distributed deep learning training protocol which provides cost-efficient and faster training of deep neural networks by utilizing the compute resources in the network." (Raven Protocol WP, p. 4)
Long-Range Vision: "Raven Protocol is the central hub for performing AI training and development. The platform will integrate AI services such as data exchanges, data labeling/annotation services, shared models, and implemented algorithms that may be developed by other projects or members of the AI community....With the correct incentivization in place, AI training can be accessible to the masses either for free or at deep discounted rates; this would undoubtedly push AI technology to levels unimaginable at present." (Raven Protocol WP, p. 19)
Issues for Follow Up (based on the Whitepaper)
- Proof of calculation: need to better understand this "primary guideline for the regulation and distribution of incentives to the compute nodes in the network." (Id. at pdf p. 11)
- AI-tuned stable coin for interoperability: "Further in the future, Ravel Protocol will incorporate additional features that will involve more stakeholders such as a dataset marketplace or a data labeling and annotation marketplace....Critical to such an interoperability is the data and model format that each AI-service expects. Raven will create the layer to translate the interoperability seamlessly, as well as passthrough the costs associated with using each service. Meaning that, we will need an AI-stable coin developed at the back-end, in partnership with larger AI projects."
Is this necessary? Why can't we use any other stable coin? The project also contemplates using USD payments; is that merely a bridge to a stable coin? - Explicit reliance on USD: "To provide a fair ground for the transactions, various costs inside the Raven Ecosystem will be pegged to USD, which will provide transparency to the customers." (Id. at pdf p. 16)
- Compute acceptance will be permissioned: "Those participating by lending their systems as compute nodes for training models are requested explicit permission by Raven. Such awareness is created among users to differ from companies that have, in the past, taken advantage of user-ignorance. We have clear focus in being transparent and explicit on how much of the user's resources will be utilized by our network." (Id. at pdf p. 11)
The Details.
