Meta has purchased a large number of NVIDIA H100 GPUs to build a AI system
According to a report by The Wall Street Journal, Meta is poised to embark on the creation of a vast natural language processing model, rivaling the prowess of OpenAI’s GPT-4, with development slated for early 2024.
Meta is gearing up by procuring a multitude of NVIDIA H100 GPUs and laying the groundwork for an expansive data center, ensuring it can adequately underpin the computational demands of such an ambitious language model.
Insider information gleaned from The Wall Street Journal suggests that Meta’s forthcoming natural language model will eclipse the performance of their previously launched commercial variant, Llama 2. It’s anticipated to excel in crafting intricate texts and conducting sophisticated content analyses.
However, for the time being, Meta remains reticent on this front. Their current focus lies predominantly on Llama 2 as they vie for dominance in the arena of vast natural language models against industry titans like Microsoft and Google.