How Meta and Microsoft are Democratizing Generative AI with Llama 2

Meta and Microsoft have a long history of collaboration on artificial intelligence (AI) and empowering developers and organizations to build impactful and innovative AI-powered solutions. At the recent Microsoft Inspire event, they announced a new milestone in their partnership: the launch of Llama 2, Meta’s open-surcharge language model, on Azure and Windows.

What is Llama 2 and what can it do?

Llama 2 is a family of AI models that can generate text and code in response to prompts, similar to other chatbot-like systems such as OpenAI’s ChatGPT and Google’s PaLM. However, unlike these closed-source models, which require a subscription or license to use, Llama 2 is open-source and free for research and commercial use. This means that anyone can download, fine-tune, deploy, and experiment with Llama 2 without any restrictions or costs.

Llama 2 is also more advanced than its predecessor, Llama 1, which was only available by request and later leaked online. Llama 2 has been trained on 2 trillion tokens from publicly available online data sources, mostly in English and focusing on factual texts. It has also been fine-tuned on over 1 million human annotations to improve its conversational skills. Moreover, it has double the context length of Llama 1, meaning it can generate longer and more coherent texts.

Llama 2 also outperforms other open-source language models on many external benchmarks, including reasoning, coding, proficiency, and knowledge tests. For example, it can generate code snippets in various programming languages, answer trivia questions, write summaries and essays, and even create poems and stories.

How can you use Llama 2?

Llama 2 is designed to enable developers and organizations to build generative AI-powered tools and experiences, such as chatbots, content creation, code generation, summarization, and more. Using your own data, you can also fine-tune Llama 2 for your specific domain or task. For instance, you can train Llama 2 to generate text in different languages, styles, or tones, or to handle specific scenarios or queries.

To get started with Llama 2, you can visit Meta’s website and fill out a download form. You can also read Meta’s research paper or technical overview for more details. Once you have downloaded the model, you can use it on Azure or Windows.

Azure is the platform for the most widely adopted frontier and open models, offering powerful tooling for model training, fine-tuning, inference, and safety. Azure customers can fine-tune and deploy the Llama 2 models easily and safely on Azure using Azure Machine Learning or Azure Cognitive Services.

Windows is the best place for developers to build AI experiences tailored to their customers’ needs. Windows developers can use Llama 2 by targeting the DirectML execution provider through the ONNX Runtime, allowing a seamless workflow as they bring generative AI experiences to their applications. Windows also offers world-class tools such as Windows Subsystem for Linux (WSL), Windows terminal, Microsoft Visual Studio, and VS Code.

Why is an open approach to AI important?

Meta and Microsoft share a commitment to democratizing AI and its benefits. They believe that an open approach to AI is the right one for the development of today’s AI models, especially those in the generative space where technology is rapidly advancing. By making AI models available openly, they can benefit everyone.

An open approach to AI brings several benefits:

• It fosters transparency and collaboration between academic partners doing LLM research. Meta has established an Open Innovation AI Research Community to promote this goal.
• It enables developers and organizations to access state-of-the-art AI models without any barriers or costs. This can spur innovation and creativity across various domains and industries.
• It allows users to have more control over their data and privacy. Users can choose which models they want to use and how they want to use them.
• It invites scrutiny and feedback from the broader AI community. This can help identify potential issues or biases in the models and improve their quality and safety.
• It builds trust and confidence in AI technologies among the public. This can help address some of the ethical and social concerns around AI.

Meta and Microsoft are excited about the possibilities that Llama 2 opens up for developers, organizations, and society at large. They invite you to join them in exploring this new frontier of generative AI.