New Delhi: Tech giants Meta and Microsoft have joined forces to introduce Llama 2, an open-source large language model developed by Meta, which will be integrated into Microsoft’s Windows operating system and Azure cloud computing platform. The collaboration was officially announced on July 18, with Meta confirming that Llama 2 is freely available for both research and commercial usage and has been optimized to operate seamlessly on Windows.
This partnership emerged in response to previous speculations surrounding Llama 2, which suggested that it was specifically designed for businesses and researchers, enabling them to create applications using Meta’s advanced AI technology stack. Meta asserted that Llama 2 has undergone training with 40% more publicly available online data sources and possesses double the contextual processing capacity of its predecessor, Llama 1. Notably, Meta claims that Llama 2 surpasses numerous competitors of open-source large language models (LLMs) in terms of coding efficiency, proficiency, reasoning, and performance on knowledge tests. However, Meta acknowledges that Llama 2 falls slightly short in comparison to closed-source competitors such as OpenAI’s GPT-4, as mentioned in one of their research papers.
Meta CEO Mark Zuckerberg expressed his enthusiasm for Llama 2 in a July 18 Instagram post, asserting that it provides researchers and businesses with access to their next-generation large language model, serving as a strong foundation for their work. Meta stated that they were astonished by the overwhelming demand for Llama 1 following its limited release in February, receiving over 100,000 requests for access. Unfortunately, the model was subsequently leaked online by a user of the imageboard website 4chan.
In terms of AI investment, Microsoft now supports two major players in the AI industry, having invested a total of $13 million in OpenAI throughout 2023 according to Fortune’s January report. Additionally, Meta recently made Llama 2 commercially available on Amazon SageMaker JumpStart, which provides developers with a comprehensive suite of machine-learning tools.
Meta’s decision to open-source Llama was met with criticism by two United States senators in June, who argued that the initial version’s “seemingly minimal” security measures could potentially enable malicious users to engage in criminal activities.
