Mark Zuckerberg has revealed that Meta’s AI team is currently training the Llama 3 model on a large-scale computer infrastructure. This infrastructure will include 350,000 NVIDIA H100 GPUs by the end of 2024. If we take into account other GPUs, the total number will reach 600,000, making it a significant advancement.
Currently, Meta has two AI development teams – Fundamental AI Research (FAIR), which focuses on research, and GenAI, which was established earlier in 2023 to enhance AI features for services like WhatsApp, Messenger, and Instagram.
Zuckerberg emphasizes Meta’s goal of creating artificial general intelligence (AGI) that is open-source and accessible to everyone. He also mentions the integration of AI services with Meta’s smart glasses project, such as the Ray Ban Meta smart glass.
In summary, Mark Zuckerberg and his team at Meta are working diligently to train the Llama 3 model on a massive computer infrastructure. Their aim is to develop AGI and make it accessible to the public while also integrating AI services with their smart glasses project. This ambitious endeavor showcases Meta’s commitment to pushing technological boundaries.
TLDR: Meta’s AI team, led by Mark Zuckerberg, is training the Llama 3 model on a large-scale computer infrastructure. Their goal is to create AGI that is open-source and accessible to all. Additionally, they plan to integrate AI services with their smart glasses project. This showcases Meta’s commitment to innovation and pushing technological boundaries.
Leave a Comment