U.S. tech giants Meta and Microsoft have recently unveiled a new edition of Meta’s open-source large language model (LLM), Llama 2, which is free to download for research and commercial use.
In a company blog post, Meta said it has included model weights and starting code for the pre-trained model, along with fine-tuned versions for conversational use, thereby broadening its venture into generative AI.
“Today’s expansion of our model catalog with Llama 2 and our partnership with Meta is a big step forward in achieving a responsible, open approach to AI,” Microsoft noted in a separate statement announcing the Llama 2 launch.
From Tuesday, Llama 2 has been included in Microsoft's Azure AI model catalog, which means developers can use it along with their cloud-native tools to enable content filtering and safety features.
Additionally, the model is fine-tuned to execute locally on Windows, offering developers a smooth workflow as they deliver generative AI experiences on diverse platforms.
Llama 2 is accessible through multiple platforms, including Amazon Web Services (AWS), Hugging Face, and others.
With this expanded partnership, Microsoft and Meta noted they are supporting an open approach to increase access to AI technologies that are expected to boost business activities worldwide.
Emphasizing its pledge to responsible building, Metais also offering a range of resources, including a user guide incorporating the latest research on responsible generative AI, to support Llama 2 users.
In addition, the Facebook parent company has implemented a policy that bans certain uses of its models.
Llama 2 is available for download for free on Meta’s dedicated website, allowing users to choose among three different model sizes – 7 billion, 13 billion, or 70 billion parameters.
Experts suggest that the capabilities of Llama 2 are similar to those of OpenAI’s free version of ChatGPT-3.5.
However, Llama 2 can be downloaded and used on any PC with a graphic card powerful enough to run the model, removing risks of data leaks.
-author-avatar_thumbnail.jpg){kind=avatar}