<aside> <img src="/icons/checkmark_gray.svg" alt="/icons/checkmark_gray.svg" width="40px" /> The goal is to train a 10-trillion parameter LLM-like model.

</aside>

The anchor goal is broken down into 2 parts;

1. Facilitate decentralized training over internet-grade connections in a way that supports 10 trillion param model.
2. Get enough active resources in the swarm that the run can take place. This requires;
   1. Technical progress; incentives for value accrual must be implemented.
      1. Protocol Models (Unextractibility)
      2. Compute verification
      3. Token based model inference
   2. Marketing. We must have a large number of active participants in the swarm.
      1. 1M active nodes