The purpose of Soma is to acquire knowledge by following curiosity.
Soma is an incentive system designed to continuously train an ultra large neural network and reward data that helps it learn.
Similar to how Bitcoin miners compete to earn block rewards, data submitted to Soma competes for rewards based on how learnable it is. The scoring of data is performed by the neural network. As the network learns, it becomes better at self critiquing the quality of data.
The fly wheel of incentivizing data, evaluating whether it is useful, and then training on the most useful examples is a novel approach that only works when the training process is open.
Great care has been taken to make Soma permission-less.
This means that anyone can contribute training data or change parts of the network architecture. Allowing anyone to “alter” the network is a radical departure from large AI labs. Instead of a top down approach, Soma uses a bottom up approach: start with simple rules and allow self assembly and the emergence of intelligence.
We believe this is more in line with how nature works, and how data is created in the world.
Quick Links
Learn more about S❍MA
Learn about core concepts, the organization of the network, and how participants are evaluated.
Get started using the network
Setup an account, and take your first steps with the network.
Explore performance
Check out the network stats, and benchmarks.
Ways to Participate
S❍MA is built and maintained by a global community of researchers, developers, and data scientists. There are several ways to get involved:
- Advance Research: Contribute to the theoretical foundations and methodologies that power SOMA's pattern recognition and knowledge synthesis
- Submit Data: Help expand the network's understanding by contributing data
- Develop Models: Create and improve the encoders that extract patterns from data
- Build Applications: Leverage the network's intelligence to create new tools and services
- Run a Node: Participate in consensus by operating network infrastructure
See the Contributing Guide and Research Forum to learn more about each path.