Awesome Green AI 🤖🌱

A curated list of awesome Green AI resources and tools to reduce the environmental impacts of using and deploying AI.

Generated with Stable Diffusion v2

In 2020, Information and Communications Technology (ICT) sector carbon footprint was estimated to be between 2.1-3.9% of total global greenhouse gas emissions. The ICT sector continues to grow and now dominates other industries. It is estimated that the carbon footprint will double to 6-8% by 2025. For ICT sector to remain compliant with the Paris Agreement, the industry must reduce by 45% its GHG emissions from 2020 to 2030 and reach net zero by 2050 (Freitag et al., 2021).

AI is one of the fastest growing sectors, disrupting many other industries (AI Market Size Report, 2022). It therefore has an important role to play in reducing carbon footprint. The impacts of ICT, and therefore AI, are not limited to GHG emissions and electricity consumption. We need to take into account all major impacts (abiotic resource depletion, primary energy consumption, water usage, etc.) using Life Cycle Assessment (LCA) (Arushanyan et al., 2013).

AI sobriety not only means optimizing energy consumption and reducing impacts, but also includes studies on indirect impacts and rebound effects that can negate all efforts to reduce the environmental footprint (Willenbacher et al. 2021). It is therefore imperative to consider the use of AI before launching a project in order to avoid indirect impacts and rebound effects later on.

All contributions are welcome. Add links through pull requests or create an issue to start a discussion.

🛠 Tools

Code-Based Tools

Tools to measure and compute environmental impacts of AI.

• CodeCarbon – Track emissions from Compute and recommend ways to reduce their impact on the environment.
  
• carbontracker – Track and predict the energy consumption and carbon footprint of training deep learning models.
  
• Eco2AI – A python library which accumulates statistics about power consumption and CO2 emission during running code.
  
• Zeus – A framework for deep learning energy measurement and optimization.
  
• Tracarbon – Tracks your device's energy consumption and calculates your carbon emissions using your location.
  
• AIPowerMeter – Easily monitor energy usage of machine learning programs.
  
• EcoLogits – Estimates the energy consumption and environmental footprint of LLM inference through APIs.
  
• GATorch – An Energy-Aware PyTorch Extension.
  

☠️ No longer maintained:

• carbonai – Python package to monitor the power consumption of any algorithm.
  
• experiment-impact-tracker – A simple drop-in method to track energy usage, carbon emissions, and compute utilization of your system.
  
• GPU Meter – Power Consumption Meter for NVIDIA GPUs.
  
• PyJoules – A Python library to capture the energy consumption of code snippets.
  

Monitoring Tools

Tools to monitor power consumption and environmental impacts.

• Scaphandre – A metrology agent dedicated to electrical power consumption metrics.
  
• PowerJoular – Monitor power consumption of multiple platforms and processes.
  
• Boagent – Local API and monitoring agent focussed on environmental impacts of the host.
  
• vJoule – A tool to estimate the energy consumption of your processes.
  
• jupyter-power-usage – Jupyter extension to display CPU and GPU power usage and carbon emissions.
  

Optimization Tools

Tools to optimize energy consumption or environmental impacts.

• Zeus – A framework for deep learning energy measurement and optimization.
  
• GEOPM – A framework to enable efficient power management and performance optimizations.
  

Calculation Tools

Tools to estimate environmental impacts of algorithms, models and compute resources.

• Boaviztapi - Multi-criteria impacts of compute resources taking into account manufacturing and usage.
• Datavizta - Compute resources data explorer not limited to AI.
• EcoDiag - Compute carbon footprint of IT resources taking into account manufactuing and usage (🇫🇷 only).
• GenAI Carbon Footprint - A tool to estimate energy use (kWh) and carbon emissions (gCO2eq) from LLM usage.
• Green Algorithms - A tool to easily estimate the carbon footprint of a project.
• MLCarbon - End-to-end carbon footprint modeling tool.
• ML CO2 Impact - Compute model emissions and add the results to your paper with our generated latex template.

📄 Papers

• Energy and Policy Considerations for Deep Learning in NLP - Strubell et al. (2019)
• Quantifying the Carbon Emissions of Machine Learning - Lacoste et al. (2019)
• Carbontracker: Tracking and Predicting the Carbon Footprint of Training Deep Learning Models - Anthony et al. (2020)
• Green AI - Schwartz et al. (2020)
• Carbon Emissions and Large Neural Network Training - Patterson, et al. (2021)
• Green Algorithms: Quantifying the Carbon Footprint of Computation - Lannelongue et al. (2021)
• Aligning artificial intelligence with climate change mitigation - Kaack et al. (2021)
• A Pratical Guide to Quantifying Carbon Emissions for Machine Learning researchers and practitioners - Ligozat et al. (2021)
• Evaluating the carbon footprint of NLP methods: a survey and analysis of existing tools - Bannour et al.(2021)
• A Survey on Green Deep Learning - Xu et al. (2021)
• Unraveling the Hidden Environmental Impacts of AI Solutions for Environment Life Cycle Assessment of AI Solutions - Ligozat et al. (2022)
• Measuring the Carbon Intensity of AI in Cloud Instances - Dodge et al. (2022)
• Estimating the Carbon Footprint of BLOOM a 176B Parameter Language Model - Luccioni et al. (2022)
• Bridging Fairness and Environmental Sustainability in Natural Language Processing - Hessenthaler et al. (2022)
• Eco2AI: carbon emissions tracking of machine learning models as the first step towards sustainable AI - Budennyy et al. (2022)
• Environmental assessment of projects involving AI methods - Lefèvre et al. (2022)
• Sustainable AI: Environmental Implications, Challenges and Opportunities - Wu et al. (2022)
• The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink - Patterson et al. (2022)
• Towards the Systematic Reporting of the Energy and Carbon Footprints of Machine Learning - Henderson et al. (2022)
• Towards Sustainable Artificial Intelligence: An Overview of Environmental Protection Uses and Issues - Pachot et al. (2022)
• Method and evaluations of the effective gain of artificial intelligence models for reducing CO2 emissions - Delanoë et al. (2023)
• Counting Carbon: A Survey of Factors Influencing the Emissions of Machine Learning - Luccioni et al. (2023)
• Making AI Less "Thirsty": Uncovering and Addressing the Secret Water Footprint of AI Models - Li et al. (2023)
• Zeus: Understanding and Optimizing GPU Energy Consumption of DNN Training - You et al. (2023)
• Chasing Low-Carbon Electricity for Practical and Sustainable DNN Training - Yang et al. (2023)
• LLMCarbon: Modeling the End-To-End Carbon Footprint of Large Language Models - Faiz et al. (2023)
• Power Hungry Processing: Watts Driving the Cost of AI Deployment? - Luccioni et al. (2023)
• Perseus: Removing Energy Bloat from Large Model Training - Chung et al. (2023)
• Timeshifting strategies for carbon-efficient long-running large language model training - Jagannadharao et al. (2023)
• Estimating the environmental impact of Generative-AI services using an LCA-based methodology - Berthelot et al. (2024)
• Towards Greener LLMs: Bringing Energy-Efficiency to the Forefront of LLM Inference - Stojkovic et al. (2024)
• Green AI: Exploring Carbon Footprints, Mitigation Strategies, and Trade Offs in Large Language Model Training - Liu et al. (2024)