Shrinking deep learning’s carbon footprint

OpenAI unveiled GPT-3, a text-generating tool that can write creative fiction, translate legalese into plain English, and respond to obscure trivia questions, the world’s largest language model, in June.

Deep learning, a type of machine learning that is modeled after the way neurons in the brain process and store information, is responsible for this most recent feat of intelligence.However, it had a significant cost:a minimum of $4.6 million and 355 years’ worth of computing time, assuming the model was trained on a standard GPU.

The primary reason for the model’s high price is its enormous size, which is 1,000 times greater than that of a typical language model.