This is the base model weights and network architecture of Grok-1, xAI's large language model. Grok-1 is a 314 billion parameter Mixture-of-Experts model trained from scratch by xAI.