Meta is making its Llama 2 key language model open source, Facebook’s parent company announced on July 18. The update to the model, which is slated to be released in February 2023 as the first-generation LLAMA (also styled as the Llama 1), was first revealed at the Microsoft Inspire event. Microsoft has become a preferred partner with META on the Llama 2.
What is Llama 2?
Llama 2 is a large language model that can be used to build generative and conversational AI models. Simply put, Llama 2, like GPT-4, can be used to build chatbots and AI assistants for commercial or research purposes.
It runs on a collection of pre-trained and refined generative text models on a scale of 7 billion to 70 billion parameters, and 2 trillion tokens of data from publicly available sources went into pre-training. Overall, this is 40% more than the tokens used to train the original Llama.
To see: Rental Kit: Fast Engineer (Tech Republic Premium)
Where is Llama 2 available?
Llama 2 can be downloaded here for Meta’s research and commercial use. Available open source resources include model weights and starting code for pre-trained models, as well as fine-tuned versions of conversational AI.
“Access to today’s AI models means a generation of developers and researchers can test them as a community and quickly identify and fix problems,” Meta wrote in a blog post about Llama 2.
Developers who already have an account with Microsoft’s Azure AI Model Catalog can access Llama 2 from there. It can be found on Amazon Web Services, Hugging Face, and other AI marketplaces. AWS customers should find it on SageMaker, the machine learning marketplace.
“Meta’s announcement that the model will be available in AWS and Microsoft Azure is a big step forward for them and reflects their ambition to become an enterprise player in the generative AI space,” Gartner analyst Arun Chandrasekaran said in an email to TechRepublic.
Meta teams up with Qualcomm for on-device AI
Qualcomm will install Lama 2 on select devices in 2024. The exact device models this will apply to is yet to be revealed, but Qualcomm has said that these will be devices powered by Snapdragon processors. Qualcomm aims to run the language model directly on some devices, not always in the cloud.
“We appreciate Meta’s vision for open and accountable AI and are committed to fostering innovation and lowering barriers to entry for developers of all sizes by bringing generative AI into devices,” said Durga Malladi, senior vice president and general manager of technology, planning and edge solutions, Qualcomm, in a press release. “To effectively mainstream generative AI, AI will need to run in the cloud as well as on edge devices such as smartphones, laptops, vehicles and IoT devices.”
What does Llama 2 say about the competition in the generic AI business sector?
Opening Llama 2 and ending the partnership with Microsoft may indicate that Meta is trying to stay competitive with GPT-4. OpenAI’s GPT-4 is also free to use and is the model behind ChatGPT, which Microsoft has bet heavily on. Google also has a horse with the PaLM model behind Bard.
“This is going to change the landscape of the LLM market,” said Yann Lecun, Meta’s chief AI scientist. on Twitter,
How the connection with Meta might change Microsoft’s deals with OpenAI is unclear at this point, but “a partnership with Meta could open up new opportunities for them,” Chandrasekaran said. “The Llama 2 model has the potential to drive demand for Azure’s IaaS and operational tools as customers look to refine these models and build business applications on top of them,” he said.
Making the model open source could also be a big change. Chandrasekaran said, “By releasing Llama 2 and licensing it for commercial use, META can provide a huge boost to the open source community.” “Today, closed source models certainly have a performance advantage over open source models, but Llama has the potential to close that gap over the medium to long term.”
Stay Connected With Us On Social Media Platforms For Instant Updates, Click Here To Join Us Facebook