Our large language AI models (LLMs) are designed to add value to Meta's applications and technology, but they are also open-sourced so that other businesses can gain from them. Large language models from our open-source collection, Meta Llama, are already being used by companies in the fields of research, medical, education, and customer service. Over 170 million downloads of our Llama models have been made. These are some illustrations of what can be done with Llama 2 and 3.
Building a Regionalized Math Education Platform using Llama
STEM education has a big impact on economic mobility and innovation. In light of this, Mathpresso, a South Korean education tech startup, and QANDA, a provider of tailored math learning materials in 50 countries, used our Llama 2 model to create MathGPT, an LLM with robust, accurate math skills.
According to Jake Yongjae Lee, co-founder of Mathpresso, "commercial LLMs like ChatGPT lack the adaptability needed for the complicated education scene."
Lee claims that hyperlocal elements like curriculum, school district, exam trends, teaching style, and more have an impact on students' learning. By overcoming this complexity, Llama 2 is assisting Mathpresso in building a society in which more individuals have access to high-quality education. The group used their own unique data and methods to produce adaptable, locally relevant instructional products using Llama 2.
Earlier this year, MathGPT set a new world record in benchmarks assessing mathematical performance at both the primary and secondary school level.
An Artificial Intelligence Friend for Zoom Business Services and Workplace
Zoom developed an LLM that enables Zoom AI Companion, their generative AI assistant, by utilizing Llama 2 and other third-party models. If you have to leave your workstation, AI Companion can bring you up to date on the proceedings of a meeting. Additionally, it may summarize Zoom Meetings, emphasize key insights and next steps, and offer beneficial presentation advice for your upcoming gathering. In Zoom Team Chat, AI Companion can also provide you with a summary of your unread messages and assist you in crafting responses that are the right length and tone.
Modifying Llama to Deliver Health Information in Low-Resource Environments
In the field of medicine, timely access to pertinent information can make all the difference.
Llama 2 was modified by researchers at Yale School of Medicine and EPFL's School of Computer and Communication Sciences to condense medical data into an understandable, conversational interface. The outcome is a collection of big multimodal foundation models called Meditron, which are intended to support clinical diagnosis and decision-making. By collaborating closely with humanitarian organizations such as the International Committee of the Red Cross (ICRC), Meditron can help medical professionals diagnose and treat patients in underserved areas and provide fast and accurate medical response in emergency situations.
The team just released Meta Llama 3, and in less than a day, they improved the new model to produce Llama-3[8B]-MeditronV1.0. When it comes to answering questions on biological tests, for example, Meditron does exceptionally well compared to industry standards.
According to Mary-Anne Hartley, a professor at Yale University and co-leader of the initiative, "foundation models have become modern-day intellectual and cultural treasures." They have the capacity to offer counsel and direction that could save lives when used in the medical field. However, the settings with the fewest resources stand to gain the most and continue to be underrepresented.
The idea behind open models like Meditron is that they can facilitate fair access to medical information.
These are just a few instances of how businesses are utilizing Meta Llama to accomplish their objectives and improve the lives of people. We will keep sharing the useful ways that people are utilizing Meta Llama.