Microsoft and Meta Deepen Partnership to Boost AI Capabilities with New Gen Llama

Microsoft and Meta, formerly Facebook Inc., have announced the launch of their next generation of large language models (LLMs). Llama 2 will be the latest addition to their growing family of LLMs on Azure and Windows. The collaboration will deepen Microsoft and Meta’s existing alliance. Meta identified Microsoft as their preferred partner with Llama 2. Currently, Llama 2 is accessible to businesses and researchers. They intend to allow access to a large cohort of businesses. Customers will also include academics, tech industry experts, and anyone else who recognizes the value of accelerating AI technology advancements. It is currently under public review.

Llama 2’s capabilities will offer developers flexibility in the type of model they wish to create and support open and experimental models. The approach aims to make it possible for organizations and developers to create generative AI tools and experiences. Meta and Microsoft are dedicated to liberalizing AI and its benefits.

John Montgomery, Corporate Vice President, Azure AI at Microsoft, in the announcement stated,

The announcement builds on our partnership to accelerate innovation in the era of AI and further extends Microsoft’s open model ecosystem and position as the world’s supercomputing platform for AI.

Democratizing AI through the power of partnership.🎉 We're excited to welcome Llama 2 from @Meta to @Azure and @Windows: https://t.co/OJyYP9sVBA

— Microsoft (@Microsoft) July 18, 2023

Customers will now be able to modify and implement Llama 2’s 7B, 13B, and 70B parameters in a more direct and safe manner on Azure. The model will also support Windows PCs. Developers can use Llama by focusing on DirectML execution source through ONNX Runtime. In addition to integrating AI into their apps, it will produce a seamless workflow.

Meta and Microsoft’s evolving relationship

For AI development, Meta and Microsoft have a longstanding relationship. The collaboration first began with the integration of ONNX Runtime with PyTorch to refine the developer experience and Azure as Meta’s choice for a strategic cloud provider.

Azure’s specially designed AI supercomputing platform is created to assist the world’s top AI organizations in developing, honing, and utilizing some of the most resilient AI capabilities. The Llama 2 model equips programmers with tools to use Azure AI’s robust modeling, training, modifying, interpreting, and supporting abilities. It will be the latest addition to Microsoft’s Azure AI model catalog. This catalog will act as a central point for foundation models. This will enable developers and machine learning (ML) experts to find, examine, customize, and distribute huge pre-built AI models widely.

A Vow to Responsibility

Both companies pledge to build AI that is centered around transparency and access. They know the risks that accompany AI. Meta and Microsoft are committed to building responsible AI models and provide a number of resources to help those who use Llama 2

• Safety testing: Several internal and external efforts have red-teamed or safely tested the finely refined models. To help improve the model, their teams created an aggressive prompt. They also commissioned third parties to conduct external combat testing to find gaps in performance.

• Transparency: Both companies promise to outline the model’s tuning and evaluation procedures as well as point out any areas of improvement. Their transparency plan reveals known difficulties and problems they have encountered, offering suggestions to resolve them.

• Responsible resource for developers: In order to help developers with responsible development and safety assessments, they have also created a user guide. These procedures discuss the best practices for the best research underway right now on ethical generative AI.

• Suitable Use Policy: To ensure these models are used ethically and responsibly, policies have been implemented prohibiting specific use cases.

Llama 2 will help programmers build customized experiences via a GitHub repo. They can fine-tune LLMs to meet their specific needs on Windows PCs using Subsystem for Linux and highly capable GPUs. Both Meta and Microsoft agree that making LLMs publicly accessible will aid in the creation of useful and secure generative AI

We are eager to witness what innovations and advancements this partnership brings in order to revolutionize the realm of AI!

Read More: KPMG-Microsoft Amplify Partnership To Unlock New AI Possibilities