REVIEW 25 cited by
Carbontracker: Tracking and Predicting the Carbon Footprint of Training Deep Learning Models
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
Deep learning (DL) can achieve impressive results across a wide variety of tasks, but this often comes at the cost of training models for extensive periods on specialized hardware accelerators. This energy-intensive workload has seen immense growth in recent years. Machine learning (ML) may become a significant contributor to climate change if this exponential trend continues. If practitioners are aware of their energy and carbon footprint, then they may actively take steps to reduce it whenever possible. In this work, we present Carbontracker, a tool for tracking and predicting the energy and carbon footprint of training DL models. We propose that energy and carbon footprint of model development and training is reported alongside performance metrics using tools like Carbontracker. We hope this will promote responsible computing in ML and encourage research into energy-efficient deep neural networks.
Forward citations
Cited by 25 Pith papers
-
Towards Sustainability Model Cards
A YAML-based DSL and validating parser, Sustainability Model Cards, formalizes ML sustainability reporting across training, inference, and platform dimensions, extending the Model Cards initiative.
-
Towards Decentralized and Sustainable Foundation Model Training with the Edge
Idle edge devices could cut foundation-model training carbon by an estimated 4-8x versus a cloud GPU, based on small-scale energy measurements and lifecycle accounting.
-
Non-Markovianity and memory enhancement in Quantum Reservoir Computing
Non-Markovian quantum reservoir computers can revive memory of past inputs, while Markovian reservoirs are claimed to obey an exponential memory decay bound.
-
Towards Sustainable NLP: Insights from Benchmarking Inference Energy in Large Language Models
A broad benchmark shows LLM inference energy scales with output length and response time, while batch size, quantization, and prompt phrasing can reduce it.
-
How Green Can AI Be? A Study of Trends in Machine Learning Environmental Impacts
Training large AI models has become exponentially more energy-intensive and environmentally impactful since 2012, and efficiency gains and greener electricity alone cannot offset this growth.
-
Language Models in Software Development Tasks: An Experimental Analysis of Energy and Accuracy
Larger language models do not reliably deliver higher accuracy on software tasks, and quantized large models can outperform full-precision medium models on energy and accuracy together.
-
Is Locational Marginal Price All You Need for Locational Marginal Emission?
Under a uniqueness assumption, LME is a unique function of LMP, computable from pre-computed critical regions of the security-constrained economic dispatch problem.
-
Robustness of transferability estimation metrics for medical imaging
Transferability estimation metric rankings in medical imaging are unstable to target resampling and to the evaluation metric used for the reference ranking.
-
Keyword Matters: Unveiling the Energy Sensitivity of On-Device LLM Prompting
On-device LLMs use measurably different amounts of energy depending on the prompt verb, but the effect is mostly a side-effect of response length and is inconsistent across models.
-
HAFM: Hierarchical Autoregressive Foundation Model for Music Accompaniment Generation
HAFM uses a hierarchical autoregressive model with dual-rate HuBERT and EnCodec tokens to generate coherent instrumental music from vocals, achieving FAD 2.08 on MUSDB18 while matching prior systems with fewer parameters.
-
Energy and Quality of Surrogate-Assisted Search Algorithms: a First Analysis
A first empirical study of the energy consumption of surrogate-assisted particle swarm optimization, proposing energy and surrogate accuracy as evaluation axes.
-
Calculating Software's Energy Use and Carbon Emissions: A Survey of the State of Art, Challenges, and the Way Ahead
A structured survey of 21 software energy and carbon calculation tools, organized as Monitoring, Estimation, or Black-Box approaches, with a component-wise comparison and a list of open challenges.
-
Diffused Responsibility: Analyzing the Energy Consumption of Generative Text-to-Audio Diffusion Models
Inference energy of seven text-to-audio diffusion models grows linearly with denoising steps, while quality saturates, so the best quality-per-energy settings use 10 to 50 steps.
-
YINYANG-ALIGN: Benchmarking Contradictory Objectives and Proposing Multi-Objective Optimization based DPO for Text-to-Image Alignment
Introduces a six-axis contradictory-objective benchmark and a weighted DPO variant (CAO), but the claimed balanced alignment rests on evaluations that reuse the training objectives.
-
Learning in Log-Domain: Subthreshold Analog AI Accelerator Based on Stochastic Gradient Descent
The paper maps SGD with L2 regularization into continuous-time differential equations and implements them with subthreshold log-domain circuits, simulating a single node and a linear regression with under 1% error ver...
-
Spend More to Save More (SM2): An Energy-Aware Implementation of Successive Halving for Sustainable Hyperparameter Optimization
SM2 is an energy-aware successive halving optimizer that selects batch size and learning rate using GPU energy per epoch and cyclical learning-rate exploration, cutting HPO energy by 8% to 47% in three test scenarios.
-
Towards Sustainable Artificial Intelligence: A Comprehensive Review and Comparative Analysis of Deep Learning Models' Carbon Footprint
On a CPU-based CIFAR-10 benchmark, training accounts for nearly all carbon emissions, and more complex models do not yield proportionally higher accuracy.
-
A Discrepancy-Based Perspective on Dataset Condensation
Dataset condensation is reframed as minimizing distribution discrepancies, and existing methods are sorted into a taxonomy; no new algorithm or experiments are provided.
-
Performance is not All You Need: Sustainability Considerations for Algorithms
The paper introduces FMS and ASC, composite sustainability scores that fuse accuracy and energy consumption, and evaluates them on multiple vision tasks.
-
Quantifying the Energy Consumption and Carbon Emissions of LLM Inference via Simulations
A simulation framework couples an LLM inference simulator with a GPU power model and an energy-grid co-simulator to estimate energy and carbon emissions across deployment configurations.
-
Greening AI-enabled Systems with Software Engineering: A Research Agenda for Environmentally Sustainable AI Practices
A 29-participant workshop synthesized a research agenda for reducing AI's environmental footprint through software engineering, covering measurement, benchmarking, architecture, empirical methods, and education.
-
Beyond Accuracy: EcoL2 Metric for Sustainable Neural PDE Solvers
EcoL2 combines relative error with lifecycle carbon emissions into one score, and experiments show it exposes large carbon differences among neural PDE solvers with similar accuracy.
-
RESQUE: Quantifying Estimator to Task and Distribution Shift for Sustainable Model Reusability
RESQUE is a single index, computed from representation angles or cluster-label agreement, that correlates with measured retraining cost, energy, and carbon emissions across several vision models and datasets.
-
Federated Learning in Mobile Networks: A Comprehensive Case Study on Traffic Forecasting
On a Barcelona LTE dataset, federated LSTM training matches or beats centralized and per-base-station training for traffic forecasting and scores best on a hand-weighted sustainability metric.
-
Optimization Strategies for Enhancing Resource Efficiency in Transformers & Large Language Models
The paper reports energy and accuracy trade-offs for LLM compression methods, but its proposed optimization equation is marred by an internal formula error.
Discussion (0). Continue with ORCID to comment.