Estimating Deep Learning energy consumption based on model architecture and training environment

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: del Rey, Santiago, Cruz, Luís, Franch, Xavier, Martínez-Fernández, Silverio
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918147559260160
author del Rey, Santiago
Cruz, Luís
Franch, Xavier
Martínez-Fernández, Silverio
author_facet del Rey, Santiago
Cruz, Luís
Franch, Xavier
Martínez-Fernández, Silverio
contents To raise awareness of the environmental impact of deep learning (DL), many studies estimate the energy use of DL systems. However, energy estimates during DL training often rely on unverified assumptions. This work addresses that gap by investigating how model architecture and training environment affect energy consumption. We train a variety of computer vision models and collect energy consumption and accuracy metrics to analyze their trade-offs across configurations. Our results show that selecting the right model-training environment combination can reduce training energy consumption by up to 80.68% with less than 2% loss in $F_1$ score. We find a significant interaction effect between model and training environment: energy efficiency improves when GPU computational power scales with model complexity. Moreover, we demonstrate that common estimation practices, such as using FLOPs or GPU TDP, fail to capture these dynamics and can lead to substantial errors. To address these shortcomings, we propose the Stable Training Epoch Projection (STEP) and the Pre-training Regression-based Estimation (PRE) methods. Across evaluations, our methods outperform existing tools by a factor of two or more in estimation accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2307_05520
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Estimating Deep Learning energy consumption based on model architecture and training environment
del Rey, Santiago
Cruz, Luís
Franch, Xavier
Martínez-Fernández, Silverio
Machine Learning
Computers and Society
Software Engineering
D.2; I.2
To raise awareness of the environmental impact of deep learning (DL), many studies estimate the energy use of DL systems. However, energy estimates during DL training often rely on unverified assumptions. This work addresses that gap by investigating how model architecture and training environment affect energy consumption. We train a variety of computer vision models and collect energy consumption and accuracy metrics to analyze their trade-offs across configurations. Our results show that selecting the right model-training environment combination can reduce training energy consumption by up to 80.68% with less than 2% loss in $F_1$ score. We find a significant interaction effect between model and training environment: energy efficiency improves when GPU computational power scales with model complexity. Moreover, we demonstrate that common estimation practices, such as using FLOPs or GPU TDP, fail to capture these dynamics and can lead to substantial errors. To address these shortcomings, we propose the Stable Training Epoch Projection (STEP) and the Pre-training Regression-based Estimation (PRE) methods. Across evaluations, our methods outperform existing tools by a factor of two or more in estimation accuracy.
title Estimating Deep Learning energy consumption based on model architecture and training environment
topic Machine Learning
Computers and Society
Software Engineering
D.2; I.2
url https://arxiv.org/abs/2307.05520