Meta’s Open Source Llama 2 Might Shake ChatGPT
On Tuesday, Meta released their new open-source artificial intelligence language model, Llama 2, which will be free to any individual and all businesses. Llama 2, an updated version of Llama 1, is trained on 40% more data with two times the context length. Further, compared to other open-source language models, such as Falcon and MPT, Llama 2 surpasses them in multiple benchmarks, including reasoning, coding, and more.
In a statement, Meta explains their philosophy behind the recent open-source model: “We support an open innovation approach to AI. Responsible and open innovation gives us all a stake in the AI development process, bringing visibility, scrutiny and trust to these technologies. Opening today’s Llama models will let everyone benefit from this technology.”
To further encourage an innovative space, Meta is also introducing Open Innovation AI Research Community, a program designed to allow academic researchers to share their expertise with rolling acceptance till September 10, 2023, as well as a Llama Impact Challenge to support the use of AI to tackle critical environmental, education and other issues.
The widely available Llama 2 will lower the barrier of entry to the emerging market of AI software and, thus, may increase the threat of competition for established players such as OpenAi, developer of ChatGPT.
