Social media giant Meta, along with Microsoft, have announced their latest iteration in artificial intelligence (AI): Llama 2.
Meta says it believes in an open approach for developing AI models, especially within the rapidly advancing generative space. According to the company, by making AI models openly accessible, they are providing businesses, startups, researchers, and entrepreneurs with tools that could be challenging to develop independently. Such tools, supported by expansive computing power, can open up a world of opportunities for innovation and economic growth.
Additionally, the company states that open access to AI models is safer as it allows developers and researchers to conduct thorough tests, identifying and rectifying issues quickly. This collaborative approach also allows Meta’s teams to learn from the usage patterns of these tools, further improving them and rectifying any vulnerabilities.
Meta says its commitment to exploratory research, open source and collaboration with academic and industry partners, has already demonstrated the power of innovation in the open. Large language models released by the company have been driving progress for developers and researchers, serving as the core of many new AI-powered experiences. Meta says the huge demand for Llama 1 from researchers, with over 100,000 access requests, indicates the significant interest in AI development.
Meta says it now stands ready to open source the next version, Llama 2, which is available free of charge for both research and commercial use. Announced by Microsoft’s Satya Nadella at Microsoft Inspire, the partnership between the two tech giants has reached a new height with Microsoft as the preferred partner for Llama 2. The AI model is available in Microsoft’s Azure AI model catalogue and optimized to run locally on Windows, offering a seamless workflow for developers to bring generative AI experiences across different platforms. Llama 2 is also available through Amazon Web Services (AWS), Hugging Face, and other providers.
Microsoft and Meta say the partnership between the two commpanies has been mutually beneficial, introducing an open ecosystem for interchangeable AI frameworks and advancing the state of AI. Their collaboration extends beyond AI, reaching into the metaverse to deliver immersive experiences for the future of work and play.
The expanded partnership between Microsoft and Meta aims to provide increased access to foundational AI technologies, benefiting businesses worldwide. This approach is supported by a broad range of global supporters, including companies eager to build new products with Llama 2, cloud providers who will include Llama 2 in their offerings, research institutions collaborating on the deployment of large generative models, and people from tech, academia, and policy sectors who share the vision.
Meta says that, throughout its history, it has embraced an open-source approach, developing and sharing frameworks that have become industry standards, like React and PyTorch. The company believes that openly sharing large language models will similarly contribute to the development of safer generative AI.