Tag
A tweet announces the live training of a 535B parameter (23B active) large language model, with links to follow the process on Weights & Biases and GitHub.