Meta (or Facebook, as most of us still think of it) this week announced the availability of Llama 2 (Large Language Model Meta AI 2), the next generation of its open source large language model, which throws down something of a gauntlet to closed commercial models such as OpenAI. Llama 2 is free for research and commercial use. As expected, Microsoft and Meta are expanding their partnership, with Microsoft the preferred partner for Llama 2.
Llama 2 is now available in the Azure AI model catalogue, enabling developers using Microsoft Azure to build with it and leverage their cloud-native tools for content filtering and safety features. It is optimized to run locally on Windows. Llama 2 is also available through Amazon Web Services (AWS), Hugging Face, and other providers.
In a blog post announcing Llama 2, Meta said: “Opening access to today’s AI models means a generation of developers and researchers can stress test them, identifying and solving problems fast, as a community. By seeing how these tools are used by others, our own teams can learn from them, improve those tools, and fix vulnerabilities.”
It also gives people the opportunity to create alternatives to the likes of ChatGPT and Google’s Bard, diluting the competition.
Llama 2 is not as powerful as GPT: it comes in three sizes: 7 billion; 13 billion; and 70 billion parameters, which correlate with performance and accuracy. By comparison, OpenAI’s GPT 3.5 has up to 175 billion parameters (they declined to give details for GPT4). Bard has 137 billion parameters. However those models require huge computing power, while models such as Llama open the door to more accessible and resource friendly LLMs.
Commenting on the launch of Llama 2, tech publication VentureBeat said: “By launching LLaMA 2, Meta has taken a significant step in opening AI up to developers worldwide. As developers begin to customize and build upon this new model, we can expect to see a surge of innovative AI applications in the near future.”
It also, however, opens up fears around how open source LLMs will be regulated, and Meta’s “acceptable use” policy is unlikely to be enough to prevent abuse of its licensing terms.