The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink
David S. Patterson, Joseph E. Gonzalez, Urs Hölzle, Quoc V. Le, Liang Chen, Lluís-Miquel Munguía, Daniel Rothchild, David R. So, Maud Texier, Jeff Dean
Abstract
Machine learning (ML) workloads have rapidly grown, raising concerns about their carbon footprint. We show four best practices to reduce ML training energy and carbon dioxide emissions. If the whole ML field adopts best practices, we predict that by 2030, total carbon emissions from training will decline.
Topics & Concepts
Carbon footprintComputer scienceFootprintPlateau (mathematics)Training (meteorology)Raising (metalworking)Memory footprintCarbon fibersCarbon dioxideArtificial intelligenceGreenhouse gasEnvironmental economicsAlgorithmMechanical engineeringMeteorologyOperating systemEngineeringGeologyMathematicsEconomicsComposite numberEcologyOceanographyPaleontologyMathematical analysisPhysicsBiologyExplainable Artificial Intelligence (XAI)