NVIDIA CEO Hints at GPT-4's Scale with 1.8 Trillion Parameters

During the NVIDIA GTC conference, Jensen Huang, the CEO of NVIDIA, hinted at the scale of the latest OpenAI model, GPT-4, suggesting it has approximately 1.8 trillion parameters. This information, which seems to confirm the specifications of GPT-4, also referred to as GPT-MoE, has sparked discussions among technology enthusiasts and experts. The revelation, initially speculated upon, appears to have been inadvertently confirmed through Huang's statements and a presentation slide, marking a significant moment in the development of AI technologies.