After promising to open source its Grok AI chatbot, Elon Musk’s xAI has done just that as promised.
On Monday, xAI released Grok-1 on GitHub, allowing anyone to leverage the model, under the Apache 2.0 license, in the name of transparency. This means commercial use is allowed but there is not data available that Grok was trained on (mainly posts on X).
“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,” said the company.
“This is the raw base model checkpoint from the Grok-1 pre-training phase, which concluded in October 2023. This means that the model is not fine-tuned for any specific application, such as dialogue,” said xAI.
The Grok-1 large language model only uses one-fourth of its compute to ensure for a fast and quick response.
here's your DEEP DIVE into @grok's architecture!
I just went through the https://t.co/8Y5cjeImg6, for this 314B open source behemoth with *no strings attached*.— Andrew Gao (@itsandrewgao) March 17, 2024
That image above? It was created using the AI image maker Midjourney, “based on the following prompt proposed by Grok: A 3D illustration of a neural network, with transparent nodes and glowing connections, showcasing the varying weights as different thicknesses and colours of the connecting lines,” said xAI.
In order to use the Grok AI chatbot, one has to be an X Premium+ customer. But those capable of leveraging the Grok-1 source code on GitHub, you can go at it if you have the necessary powerful hardware; at this point, mainly researchers and developers are the only ones capable of testing it out.
xAI founder Musk currently is suing OpenAI, the creator of ChatGPT, alleging the company has diverged from its original goal of being a non-profit. OpenAI has rejected the claims, and the legal battle is ongoing.
