Artificial intelligence chip startup Cerebras Systems announced on Tuesday that it has released open-source ChatGPT-like models for the research community and businesses.
The company stated that the models can be used for free in an effort to promote greater cooperation.
Silicon Valley-based Cerebras has released seven models, all trained on an AI supercomputer called Andromeda, ranging from smaller language models of 111 million parameters up to one large model. more than 13 billion parameters.
“There is a big movement to close what has been open sourced in AI…it`s not surprising as there’s now huge money in it,” said Andrew Feldman, founder and CEO of Cerebras.
For instance, OpenAI’s ChatGPT chatbot, which launched late last year, has 175 billion parameters and can generate poetry and research, which has helped drive widespread interest and funding for AI.
Cerebras says smaller models can be deployed on smartphones or speakers while larger models work on PCs or servers, although complex tasks like compositing large chunks require larger models.
Feldman said his largest model took just over a week to train, a job that can often take months, thanks to the Cerebras system architecture, which consists of a chip the size of a plate designed designed to train AI.