The company is buying NVIDIA's chips to accelerate its progress.
Image credit: Skorzewiak
While OpenAI's ChatGPT is still the most advanced AI chatbot, it might soon get new competition. Meta is allegedly working on its own model that should be as powerful as GPT-4.
According to The Wall Street Journal, it is buying NVIDIA's H100 AI-training chips to build the system. The large language model (LLM) is expected to be several times more advanced than Llama 2 – Meta's latest open-source model. The company should start training the new LLM in early 2024.
Llama 2 was launched in July, set on Microsoft's Azure cloud services. NVIDIA's chips will help Meta end this dependence and train the AI internally.
Meta is far from the first company that tried to overshadow ChatGPT – everyone wants a piece of the tech understandably. Bloomberg reported in July that Apple is also working on a new ChatGPT-like tool. It has built its own framework to create LLMs and created a chatbot that the employees can use.
While the ChatGPT hype has calmed down, tech corporations don't plan to let it go, so I believe we'll get back to testing not one but several chatbots soon enough.
Find the original article here and join our 80 Level Talent platform and our Telegram channel, follow us on Instagram, Twitter, and LinkedIn, where we share breakdowns, the latest news, awesome artworks, and more.