How companies are using Meta Llama | Goal

We built our large AI language models (LLMs) not only to create value in Metas applications and technologies, but also to enable other organizations to benefit from these models by sourcing them. Today, Meta Llama, our collection of large open source language models are already being used by organizations in education, customer service, research and medicine. Our Llama models have over 170 million downloads. Here are some examples that show what’s possible with Llama 2 and 3.

Using Llama to create a localized mathematics education platform

STEM education is a critical factor in economic mobility and innovation. With this in mind, Mathpressoa South Korean educational technology startup, along with WHENwhich delivers personalized math learning content in 50 countries, used our Llama 2 model to create MathGPT, an LLM with powerful and accurate math capabilities.

Commercial LLMs like ChatGPT lack the customization needed for the complex educational landscape, said Mathpresso co-founder Jake Yongjae Lee.

According to Lee, student learning is influenced by hyperlocal factors such as curriculum, school district, exam trends, teaching style and more. Llama 2 is navigating this complexity and helping Mathpresso create a world where more people have access to quality education. With Llama 2, the team was able to create flexible and localized educational products while using their own specialized data and techniques.

Earlier this year, MathGPT set a new world record Benchmarks that assess math performance both at primary and secondary level.

An AI companion for Zoom Workplace and Zoom Business Services

Zoom leveraged Llama 2 and other third-party models to create an LLM that powers its generative AI assistant, Zoom AI Companion. AI Companion can catch you up on what happened in a meeting if you need to step away from your desk. It can also summarize Zoom meetings, highlight next steps and takeaways, and provide helpful presentation tips for your next meeting. AI Companion can also summarize your unread messages in Zoom Team Chat and help you craft replies that fit the length and tone you’re looking for.

Adapting the llama to provide medical information in resource-poor settings

In medicine, access to the right information at the right time can make all the difference.

Researchers a EPFLs Faculty of Computer Science and Communication Sciences i Yale School of Medicine adapted Llama 2 to compress medical information into an accessible conversational interface. The result is Meditron, a set of large multimodal foundation models designed to aid clinical decision-making and diagnosis. By working closely with humanitarian organizations such as the International Committee of the Red Cross (ICRC), Meditron has the potential to help health workers diagnose and treat patients in underserved areas, and address emergency scenarios that require a medical response fast and accurate.

Following our recent launch of Meta Flame 3the team adjusted the new model within 24 hours to deliver it Flame-3[8B]- Meditron V1.0. Meditron achieves solid results on key industry benchmarks in the field, such as answering biomedical exam questions.

The foundation’s models have become modern intellectual and cultural assets, says Yale professor Mary-Anne Hartley, who co-directs the project. When applied to the medical domain, they have the potential to provide life-saving advice and guidance. However, the lower resource settings have more to gain and remain the least represented.

The goal is that open models like Meditron can help create equitable access to medical knowledge.

These are just a few examples of how organizations are using Meta Llama to achieve their goals and make an impact on people’s lives. We’ll continue to share how others are using Meta Llama.