The base model of xAI’s chatbot Grok is now available on GitHub for anyone to use freely for their own projects.
The model was released under the Apache 2.0 license, which allows the commercial use of the underlying technology behind Grok. However, the release does not include the training code used for the model.
“We are releasing the base model weights and network architecture of Grok-1, our large language model. Grok-1 is a 314 billion parameter Mixture-of-Experts model trained from scratch by xAI,” the company said in a blog post.
It said the “raw” base model was from Grok’s pre-training phase, which concluded in October last year. “This means that the model is not fine-tuned for any specific application, such as dialogue,” xAI said.
Billionaire Elon Musk announced that he would open-source his ChatGPT rival last week, just days after suing OpenAI, which he helped found. Musk accused the Microsoft-backed firm and its CEO, Sam Altman, of abandoning its original nonprofit mission in favor of a money-making model.
OpenAI responded to the lawsuit by publishing Musk’s old emails, in which he appeared to advocate the for-profit setup, even suggesting the venture should be merged with Tesla so it could become its “cash cow.”
xAI released Grok last year to X Premium+ subscribers. Other companies that have open-sourced some of their AI models include Meta, Mistral, and Google.
Your email address will not be published. Required fields are markedmarked