Meta and Microsoft have collaborated to launch LLaMA 2, a set of language models that will be open source. They will allow developers and organizations to create AI-powered generative tools. During the Inspire 2023 event, Microsoft announced the model's release, emphasizing their shared commitment to democratizing AI.
Meta CEO Mark Zuckerberg provided additional details, explaining that LLaMA 2 was trained with 40% more data than its predecessor and boasts an improved architecture. The model underwent supervised refinement and feedback reinforcement learning to enhance safety and quality.
LLaMA 2 will be available for research and commercial use, and can be tested on platforms like Microsoft Azure, Amazon Web Services, and Hugging Face. The partnership aims to compete with OpenAI's ChatGPT and further advance AI development.
In February, Mark Zuckerberg introduced the LLaMA model. He stated that the new "LLM has shown promise in text production, dialogue, summarizing written material, and performing more complex tasks such as proving theorems or predicting protein structure." LLaMA is not a single system, but a "quartet of models" of different sizes. It is available in four versions: LLaMA-7B, LLaMA-13B, LLaMA-33B and LLaMA-65B, where the numbers indicate the number of parameters - 7, 13, 33 and 65 billion.
Support us
Winaero greatly relies on your support. You can help the site keep bringing you interesting and useful content and software by using these options: