It is true. It is totally wasteful and it wastes lots of water as well [0]. Even before that, there aren't any efficient methods of training, inference and fin

It is true. It is totally wasteful and it wastes lots of water as well [0]. Even before that, there aren't any efficient methods of training, inference and fine-tuning available today that are viable alternatives in the field of deep learning, especially after a decade of existing. Drew's point still stands, unchallenged. [0] https://gizmodo.com/chatgpt-ai-water-185000-gallons-training-nuclear-1850324249 https://gizmodo.com/chatgpt-ai-water-185000-gallons-training...