Meta announces 'new and improved' LLaMa 2 in partnership with Microsoft
Meta has partnered with Microsoft to launch an open-source large language model like ChatGPT. Called LLaMa 2, it was trained on 40% more data in comparison to LLaMa 1, which Meta had launched in February.

What sets it apart from its competitors like OpenAI’s ChatGPT and Google’s Bard is that it’s open source.
In a press release, Meta claims that LLaMa 2 has double the context length vs. LLaMa 1 and “outperforms” other LLMs like Falcon and MPT regarding reasoning, coding, proficiency, and knowledge tests.

By collaborating with Meta, which has always been a proponent of open-source models, Microsoft has its legs in both closed-source and open-source LLMs. Remember that Microsoft has heavily invested in OpenAI, which has a closed-source LLM ChatGPT.
This also comes as Microsoft announced that it would charge a monthly fee per user for using AI-powered features in its software like Office, Teams, and other apps.
Changing the LLM market landscape?
Available free of charge for research and commercial use, LLaMa 2 will feature on Windows and the cloud computing platform Azure, which will enable developers using it to build and leverage their “cloud-native tools for content filtering and safety features.”
Designed to enable developers to build more generative AI tools, the LLM was pre-trained on publicly available online data sources. Meta says the Llama-2-chat leverages publicly available instruction datasets and over 1 million human annotations.
Listing its wide array of partners and supporters like Spotify, AWS, Dropbox, IBM, Nvidia, Hugging Face, Intel, Accenture, Zoom, Meta said that LLaMa 2 will be available through Amazon Web Services (AWS), Hugging Face, and other providers.
“Recent breakthroughs in AI, and generative AI in particular, have captured the public’s imagination and demonstrated what those developing these technologies have long known — they have the potential to help people do incredible things, create a new era of economic and social opportunities, and give individuals, creators, and businesses new ways to express themselves and connect with people,” said Meta.
Meta says that it’s committed to building responsibly, which is why it got its AI red-teamed, meaning it’s been thoroughly tested for safety several times through internal and external efforts.
The company also published a research paper laying out best practices for developing LLaMa and other safety evaluations.
This isn’t the first time that Microsoft and Meta have partnered up. In 2017, they introduced an open ecosystem for interchangeable AI frameworks and have also co-authored several research papers on the advancement of AI.
The two also joined Partnership on AI’s Responsible Practices for Synthetic Media: A Framework for Collective Action, which focused on collective action in the creation and sharing of synthetic media, said the press release.