Meta, the company behind Facebook, has unveiled its latest AI language model called Llama 2. This model, which operates at a similar level to ChatGPT, can be used freely for both commercial and research purposes. It comes in three parameter sizes: 7-billion (7B), 13-billion (13B), and 70-billion (70B), each with fine-tuned chat variants.
Developers have already made Llama 2 available via API, making it accessible to businesses and researchers alike. In benchmark tests, Llama 2 consistently outperformed other open-source models, especially in math and depth of knowledge. Despite its impressive performance, Meta’s Llama 2 falls slightly short compared to ChatGPT in some instances. However, it offers a zero-cost solution for businesses interested in implementing AI, as long as they have fewer than 700 million monthly active users.
Critics have raised concerns about the access limitations on Llama 2, arguing that it restricts certain competitors. However, many high-ranking individuals in the AI industry have expressed support for Meta’s open innovation approach. They believe that allowing everyone to benefit from Llama 2 promotes transparency, scrutiny, and trust in AI technologies.
Although Meta’s Llama 2 is not classified as open source due to access restrictions, organizations like OpenUK endorse its release as a step towards democratizing AI. This move fosters collaboration and innovation within open communities. While Meta has not released a 34-billion parameter-size model, it has committed to launching further versions of Llama 2, incorporating community feedback to enhance model safety.
Microsoft will distribute Llama 2 as part of its existing partnership with Meta, making it available through Azure AI and optimized for Windows systems. Additionally, AWS and Hugging Face will offer the model on their platforms to cater to a wide range of users in the market and open-source community. Meta has also announced a collaboration with Qualcomm to bring Llama 2 to mobile devices in 2024, enabling localized and private AI experiences.
It is worth noting that Llama 2’s predecessor, LLaMA, was initially open source for research purposes before being leaked online. Despite the model’s potential for malicious use, Meta’s release of Llama 2 aims to provide a legitimate and accessible tool for the AI community.
Meta’s Llama 2 represents a significant stride forward in democratizing AI. While it may not outperform ChatGPT in every aspect, its availability and applicability at zero cost make it an attractive option for businesses looking to adopt AI technology. By embracing open innovation, Meta hopes to drive collaboration, transparency, and trust in the AI ecosystem.
