Beyond FLOPs: Energy-Aware Knowledge Distillation for Sustainable LLMs on Code-Related Task
Authors: Enrique Barba Roque, Luis Cruz, Annibale Panichella
Published in International Symposium on Empirical Software Engineering and Measurement (ESEM) 2026, Munich, Germany, 2026
Energy-aware knowledge distillation can reduce LLM inference energy by up to 90% and memory usage by 86%, while showing that FLOPs alone is not a reliable proxy for energy consumption in software engineering tasks.
