As promised, Elon Musk has open-sourced its Grok-1 AI model that powers its Grok chatbot for premium X (formerly Twitter) users. But the catch here is that none of the training code has been made public. Additionally, only the base model for Grok has gone open-source.
xAI, the startup behind Grok-1, labels the open-source model on GitHub as the “314 billion parameter Mixture-of-Expert model.”
In a recent publication on their blog, xAI shed light on the development of their latest model, Grok-1, stating that it has not been tailored for specific uses, including conversational interactions such as OpenAI’s ChatGPT.
The organization pointed out that the training process for Grok-1 involved a unique, albeit unspecified, technological stack. Furthermore, they announced that Grok-1 operates under the Apache License 2.0, enabling its deployment across a range of commercial scenarios.
Several developers of AI-driven tools have begun discussing the integration of Grok into their offerings. Arvind Srinivas, CEO of Perplexity, shared on X that his company plans to adapt Grok specifically for conversational search purposes and will offer this enhanced capability exclusively to Pro subscribers.
Yep, thanks to @elonmusk and xAI team for open-sourcing the base model for Grok. We will fine-tune it for conversational search and optimize the inference, and bring it up for all Pro users! https://t.co/CGn6cIoivT
— Aravind Srinivas (@AravSrinivas) March 17, 2024
Musk had announced last week that Grok would be going open-source soon, but unlike the Grok chatbot on X that is able to access information on the social network, the open-source version has no such connections.
High profile AI models going open-source is nothing unheard of. Several other major companies have made their AI offerings open-source including Meta’s Llama 2, Mistral’s 7B, Falcon, and A12. Additionally, in February, Google unveiled two new models for open access, named Gemma2B and Gemma7B.
However, Musk’s AI model going open-source comes soon after his lawsuit that targeted OpenAI for no longer being “open.” He alleged that OpenAI has deviated from its original mission of building an open-source AI model for the benefit of humanity and is now simply a profit making machine for Microsoft. He claims that the idea of “Open” in the name “OpenAI” comes from open-source.
Since then he has targeted OpenAI and its CEO Sam Altman multiple times on X.
Via: Tech Crunch
