Meta and Microsoft jointly announced the launch of Llama 2, a new free and open source language model. This partnership between the two companies demonstrates Microsoft’s interest in artificial intelligence (AI) without excluding other opportunities. In addition, Qualcomm plans to integrate Llama 2 into smartphones as soon as 2024.
Llama 2, developed by Meta, is presented as the next generation of language models. It is available in open source and will be freely accessible for commercial and research use. This approach differs from that of OpenAI, which no longer offers its language models, including GPT-4, as open source. Some observers consider the development of this startup increasingly opaque.
The partnership with Microsoft makes it easier to integrate Llama 2 into Azure AI and Windows, making it easier for businesses and developers to build applications based on generative AI.
You’ll have to see Llama 2 in action to fully understand its capabilities, but Meta announces that this new model has been trained on 40% more data than Llama 1.
One of the main advantages of Llama lies in its ability to be integrated locally on a machine. Indeed, unlike ChatGPT which requires a resource-intensive cloud infrastructure, a researcher demonstrated in March that it was possible to run Llama on a simple PC.
Qualcomm has announced a partnership with Meta to natively integrate Llama 2 on PCs and smartphones equipped with Snapdragon processors from 2024. It is therefore likely that the Snapdragon 8 Gen 3 will benefit from it, as well as the first Oryon chips.
In March 2023, Meta announced major layoffs as well as a redirection of its investments towards artificial intelligence, leaving aside the immediate plan to create a metaverse.
For more details about Llama, visit the official website