Meta set to release the LLaMA AI language model amid race for AI dominance
Please note that we are not authorised to provide any investment advice. The content on this page is for information purposes only.
Meta has announced it will release a new large language model to researchers. The model is the core software used for a new artificial intelligence system. The launch comes as tech giants compete for artificial intelligence (AI) dominance. Big tech companies have been racing to integrate AI technology into their products to avoid being left behind.
Meta releases a new AI language model
The race for dominance in the AI industry commenced in November last year after Microsoft-backed OpenAI released ChatGPT. ChatGPT took the internet by storm, prompting other tech giants including Google’s parent company Alphabet and China-based Baidu, to release their AI products.
Meta has become the latest tech giant interested in the AI industry. The company has announced the development of a Large Language Model Meta AI (LLaMA). In a blog post, Meta said that LLaMA would be available through a non-commercial license to researchers, civil society, academia, and government entities.
Large language models, like the one by Meta, mine a large volume of text to summarize information and generate content. This language model can answer questions using sentences that appear like humans have written them.
Meta says this model needs “far less” computing power than the previous offerings. The model has also been trained in 20 different languages, focusing on Cyrillic and Latin alphabets.
The LLaMA model is already attracting interest across the industry. The senior software analyst at D.A. Davidson, Gil Luria, commented on Meta’s announcement saying that the tech giant was testing its generative AI capabilities to implement them into products that might be released in the future.
According to Luria, Meta did not have much experience with Generative AI. However, this sector was crucial for the future of the company. The tech giant is not alone in competing against the fast-growing AI industry.
AI gathers momentum
AI has emerged as one of the fastest-growing areas in the technology space. Tech giants are heavily investing in the AI industry, as Microsoft reportedly invested $10 billion in OpenAI. Tech giants are looking to accelerate growth following massive layoffs.
According to Meta, LLaMA could outperform its competitors. It said that a LLaMA version with 13 billion parameters was capable of outperforming GPT-3, which is a predecessor of the model used to create ChatGPT.
Meta said that the 65-billion parameter used by the LLaMA model was competitive against the Chinchilla70B and PaLM-540B by Google, which is larger than the model used by Google to promote its Bard chat-powered search.
Meta attributes the performance of LLaMA to the large quantity of “cleaner” data and “architectural improvements” within the model, which boosted the training stability. LLaMA is not the first large language model by Meta. In May last year, the company released a similar model known as OPT-175B targeting researchers.
Meta also released a language model known as Galactica that was used to write scientific articles and solve mathematical problems. However, Meta pulled down the demo for this language model after it generated incorrect responses.
