Meta released its latest large language model Llama 3 and a real-time image generator as it aims to close the gap with OpenAI in the rapidly evolving field of artificial intelligence (AI). The newly unveiled models will be integrated into Meta's virtual assistant- Meta AI- which the company boasts as the most advanced among its free-to-use counterparts.
The company said on its website, “We’ve integrated Llama 3 into Meta AI, our intelligent assistant, that expands the ways people can get things done, create and connect with Meta AI. You can see first-hand the performance of Llama 3 by using Meta AI for coding tasks and problem solving.”
Read more: Oil prices surge 3% after reported Israeli strikes on Iran: What's happening?
It added, “Whether you're developing agents, or other AI-powered applications, Llama 3 in both 8B and 70B will offer the capabilities and flexibility you need to develop your ideas.”
This comes as Meta CEO Mark Zuckerberg acknowledged competition with OpenAI while saying that Meta AI is "the most intelligent AI assistant that you can freely use."
Read more: Nothing Ear, Ear (a) priced at 11,999 and 7,999 launched; tout ChatGPT integrations
What is Llama 3?
Meta's Llama 3 is the latest in its Llama series of open-source AI models and comes in two variants: one with 8 billion parameters and second with 70 billion parameters which are the 'knowledge' which the model acquires during its training.
Read more: Ex-McKinsey employee quits ₹1.7 crore job: 'Didn't leave my desk, forgot to eat'
Llama 3: Technical specifications and training
{{/usCountry}}Meta's Llama 3 is the latest in its Llama series of open-source AI models and comes in two variants: one with 8 billion parameters and second with 70 billion parameters which are the 'knowledge' which the model acquires during its training.
Read more: Ex-McKinsey employee quits ₹1.7 crore job: 'Didn't leave my desk, forgot to eat'
Llama 3: Technical specifications and training
{{/usCountry}}Llama 3 uses a tokenizer with a vocabulary of 128,000 tokens and has grouped query attention (GQA) for both the 8B and 70B parameter models. The models are trained on sequences of 8,192 tokens, Meta said. The training data for Llama 3 consists of over 15 trillion tokens which has been sourced from publicly available data, the company said.