Meta rolls out large language model amid Big Tech’s AI push

27 Feb 2023

Image: © Tada Images/Stock.adobe.com

Meta said its AI model Llama will be available at various sizes, to give more researchers the ability to understand how large language models work.

Meta has released a new research-focused large language model, as the company joins the Big Tech rush to develop AI services.

The company said this AI model is being offered at various sizes to let researchers study the system without large amounts of infrastructure. The AI model – called Llama or Large Language Model Meta AI – will be made available at sizes ranging from 7bn to 65bn parameters.

Large language models are AI systems trained on a massive volume of text and are able to perform various tasks such as generating creative text, answering questions and solving mathematical equations.

These models have risen in popularity recently with the rapid rise of ChatGPT, the advanced chatbot by OpenAI. This rapid rise has led to an AI arms race, as companies such as Microsoft and Google have been integrating AI capabilities into their services.

Meta said large language models are one of the “clearest cases of the substantial potential benefits AI can offer”, but claimed full research access is limited due to the resources required to train and run these models.

“This restricted access has limited researchers’ ability to understand how and why these large language models work, hindering progress on efforts to improve their robustness and mitigate known issues, such as bias, toxicity and the potential for generating misinformation,” Meta said in a blog post.

The company said the smaller sizes of Llama will allow more researchers to access the model and learn more about the challenges that exist within large language models.

Meta said its latest AI model will be released under a non-commercial license on a case-by-case basis to academic researchers, including those within governments, civil societies, academia and industry research labs.

“We believe that the entire AI community – academic researchers, civil society, policymakers and industry – must work together to develop clear guidelines around responsible AI in general and responsible large language models in particular,” Meta said.

“We look forward to seeing what the community can learn – and eventually build – using Llama.”

Last November, Meta released a public demo of its science-focused model called Galactica AI. However, this public demo was taken down after three days, following intense criticism from users who said it generated biased and misleading content.

10 things you need to know direct to your inbox every weekday. Sign up for the Daily Brief, Silicon Republic’s digest of essential sci-tech news.

Leigh Mc Gowran is a journalist with Silicon Republic

editorial@siliconrepublic.com