New Delhi: Now computers will answer your every question in minutes and this luring machine will work so fast that you too will be surprised. Yes, Alphabet Inc. Google has claimed that it has built the world’s fastest Artificial Intelligence supercomputer. Google claims that it is faster and more efficient than Nvidia systems, not only that its learning model tech is also very strong.
Google has developed its own custom chip called Tensor Processing Unit (TPU). It is capable of answering over 90% of the company’s human queries on artificial AI training. It was told by the company has achieved this feat using ML models in TensorFlow, Jax, and Lingo. Four models are trained for 30 seconds. To put this in perspective, it took more than three weeks to train one of these models on the most advanced hardware accelerator available in 2015.
Google said it had built a system with more than 4,000 TPUs with custom components designed to run and train AI models. It has been running since 2020 and was used to train Google’s PaLM model, which competes with OpenAI’s GPT model, in 50 days.
