Jul 02, 2024
Meta, the tech giant behind Facebook and WhatsApp, has been rumoured to be working on its own AI model, Llama 3. However, it is a behemoth of a model, which is trained on over 400 billion parameters, making it one of the most powerful AI model integration of the world.
But to put that in perspective, Llama 3, which came out back in April 2024, was only on a meagre 70 billion parameters. So this new version is quite the upgrade, and it's already returning many of the results people expected.
Early testing has revealed that Llama 3 400B has scored as high as 86.1 on the MMLU benchmark, at the level of performance of GPT-4. Furthermore, this future innovation achieves this result with fewer than half the parameters contained in the GPT-4 model.
Read more: Apple to announce Google Gemini deal this fall
In essence, this proves that the tech giant has enhanced model architectures and training methods for Llama 3 to be a state-of-the-art AI model.
What does this mean for an average user?
This can be seen in the integration of Llama 3 into WhatsApp, which is going to provide users with a very powerful AI model right inside the messaging app. With this ability, users are enabled to pose more complex questions, get better answers, and generate more exciting content than ever.
But Llama 3's impact goes beyond just WhatsApp. It has the potential to revolutionise the way we interact with AI models. With its efficiency and power, Llama 3 could be used in a variety of applications, from customer service chatbots to language translation software.
