April 10 (Portaltic/EP) –
Goal has announced the launch of its next open source language model based on Artificial Intelligence (AI) Llama 3, that will arrive next month and that will promote different generative AI assistants throughout the company's ecosystem.
The firm led by Mark Zuckerberg announced this movement within the framework of Meta AI Day, held this Tuesday in London, thus confirming what it had announced only a day before. The Information.
The company publicly launched Llama 2 in July of the year, a model focused on research and commercial use with which it sought to provide companies with free access to tools developed on a scale “that would be difficult to build on their own,” as it explained. so.
Llama 2 consists of adjusted models with sizes of 7 billion, 13 billion and 70 billion parametersalso being a generation trained with 40 percent more data than the official version of Llama, which was not officially released.
Meta has now announced that in May it hopes to begin rolling out a set of next-generation base models, Llama 3, which will consist of “different capabilities and versions” that will be released over time. during the course of this yearaccording to Meta's president of global affairs, Nick Clegg.
For his part, Meta's product director, Chris Cox, has announced that the objective is to use Llama 3 to promote multiple products of the technology company.
With Llama 3, the company hopes to address the limitations of the previous model, while introducing new functionalities. This model will also exceed the 140,000 million parametersso you will improve your reasoning abilities.
This means that the model is expected to provide more precise answers and handle a broader spectrum of questions, including those that deal with controversial topics, as anticipated. TechCrunch.
For her part, the vice president of the AI Research area, Joelle Pineau, explained that they have the objective of “making Meta AI powered by Llama become the most useful assistant in the world”, according to this medium.