Evaluating the carbon footprint of NLP methods: a survey and analysis of existing tools

Bannour, Nesrine; Ghannay, Sahar; Névéol, Aurélie; Ligozat, Anne-Laure

doi:10.18653/v1/2021.sustainlp-1.2

Cited by 3 publications

(3 citation statements)

References 27 publications

(27 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our tool builds upon this work by making carbon tracking on cloud instances possible, enabling a larger portion of ML model training work to profit from fine-grained carbon estimation. However, recent work has found that their results vary significantly and are not fully representative of the true emissions incurred by training [3]. Perhaps most similar to our work, EnergyVis [41] is an interactive tool for visualizing and comparing energy consumption of ML models as a function of hardware and physical location (U.S. state), given metadata about a model's energy use per epoch.…”

Section: Related Workmentioning

confidence: 59%

“…For a given datacenter, this can be turned into a factor which can be multiplied against the electricity consumption of computing equipment to get an estimate of the total consumption. Some companies have highlighted particularly low PUEs, such as Google claiming a PUE of 1.10 across its fleet of data centers for the 12 months ending in Q1 2021, 3 compared to an average global PUE of 1.59 [2].…”

Section: The Scope Of Our Tool: Gpu Computation Of a Single Cloud Ins...mentioning

confidence: 99%

“…We note that our conclusions drawn from experiments and analyses on time-shifting and location-shifting are still applicable with tools that measure more electricity than just the GPU 3. https://www.google.com/about/datacenters/efficiency/4 One of the largest single source of CO 2 emissions, contributing to 7%-8% of global emissions, is the production of cement[20].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Measuring the Carbon Intensity of AI in Cloud Instances

Dodge

Taylor

Combes

et al. 2022

2022 ACM Conference on Fairness, Accountability, and Transparency

View full text Add to dashboard Buy / Rent full text

The advent of cloud computing has provided people around the world with unprecedented access to computational power and enabled rapid growth in technologies such as machine learning, the computational demands of which incur a high energy cost and a commensurate carbon footprint. As a result, recent scholarship has called for better estimates of the greenhouse gas impact of AI: data scientists today do not have easy or reliable access to measurements of this information, which precludes development of actionable tactics. We argue that cloud providers presenting information about software carbon intensity to users is a fundamental stepping stone towards minimizing emissions.In this paper, we provide a framework for measuring software carbon intensity, and propose to measure operational carbon emissions by using location-based and time-specific marginal emissions data per energy unit. We provide measurements of operational software carbon intensity for a set of modern models covering natural language processing and computer vision applications, and a wide range of model sizes, including pretraining of a 6.1 billion parameter language model. We then evaluate a suite of approaches for reducing emissions on the Microsoft Azure cloud compute platform: using cloud instances in different geographic regions, using cloud instances at different times of day, and dynamically pausing cloud instances when the marginal carbon intensity is above a certain threshold. We confirm previous results that the geographic region of the data center plays a significant role in the carbon intensity for a given cloud instance, and find that choosing an appropriate region can have the largest operational emissions reduction impact. We also present new results showing that the time of day has meaningful impact on operational software carbon intensity.Finally, we conclude with recommendations for how machine learning practitioners can use software carbon intensity information to reduce environmental impact.

show abstract

Section: Related Workmentioning

confidence: 59%

Section: The Scope Of Our Tool: Gpu Computation Of a Single Cloud Ins...mentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation