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.
This is huge. Why doesn't Grok-1 get enough votes to rank on the top? It's at least near GPT-4 level if not exceeding it.
I like to see a smaller version of this LLM that use less memory; not everyone can afford this huge memory GPU machine to run it
The open-source release of Grok-1, xAI's LLM is a boon for those interested in advanced prediction and decision-making AI models. Despite its hardware demands (like multiple H200 GPUs), this unlocks access for experimentation and the creation of groundbreaking applications.
Great hunt! I like that they open sourced it but not the reason behind open sourcing it. Outside of that, awesome to see models of this size being thrown in the public. Interested to see what people will do with it!
