We are trailblazing large and decentralized AI training and inference.
We want to build to free SOTA LLMs that beat proprietary equivalents. Free as in freedom.
We use recent research in LLMs and innovations led by the Bitcoin project to introduce a novel and scalable regime for (1) training a huge LLM from scratch and (2) performing inference over it.
We envison a future where open source models are created and used in the open as well. We will work towards this mission by training a 7B LLM on the completion of a v1 for our network. In the meantime, read our whitepaper.