Accuracy is not the only Metric that matters: Estimating the Energy Consumption of Deep Learning Models
This addresses the environmental impact of machine learning for practitioners, but it is incremental as it builds on existing concerns and methods.
The paper tackles the problem of high energy consumption and carbon footprint in deep learning models by creating an energy estimation pipeline that predicts energy needs without actual training, using collected energy data and a baseline model for layer-wise energy accumulation.
Modern machine learning models have started to consume incredible amounts of energy, thus incurring large carbon footprints (Strubell et al., 2019). To address this issue, we have created an energy estimation pipeline1, which allows practitioners to estimate the energy needs of their models in advance, without actually running or training them. We accomplished this, by collecting high-quality energy data and building a first baseline model, capable of predicting the energy consumption of DL models by accumulating their estimated layer-wise energies.