In this talk, I aim to persuade the audience of the importance of regarding energy as a first-class metric for deep learning systems, and present the current state of deep learning energy optimization with Zeus (https://ml.energy/zeus). Zeus provides convenient tools for measuring GPU time and energy consumption inside arbitrary ML applications — including large model training and serving — and optimizing energy efficiency. Finally, I'll share our vision towards making energy-efficient deep learning as easy as possible while being mindful of existing important metrics such as speed and model quality.

Jae-Won Chung is a third year PhD student at the University of Michigan. His research interests are at the intersection of software systems and deep learning, with a recent focus on energy-efficiency. He leads the ML.ENERGY (https://ml.energy) initiative. He received his Bachelor's degree in ECE from Seoul National University.