Following the development of Green AI concept, we created an open-source software tool eco2AI to help data scientists and researchers to track energy consumption and equivalent GHG emissions of their AI & ML models in a straightforward way. In eco2AI we focus on accurate tracking of energy consumption and regional carbon footprint emissions accounting. Eco2ai can help reduce emissions from AI & ML models. Only for the first 4 months, the library was used by almost 8,000 data scientists from around the world, including the RF, UK, France, Finland, the Philippines, Indonesia and others. Eco2ai directly affects the next goals: 12. Ensure sustainability consumption and production metrics 13. Take urgent action to combat climate change and its impacts. And indirectly affects goals 9 and 17.
In the paper devoted to eco2ai (https://arxiv.org/abs/2208.00406) we present examples of eco2AI usage for tracking fine-tuning of big text2image models Malevich and Kandinsky and also for optimization of GELU activation function integrated to Malevich model. eco2AI demonstrates that usage of 3-bit GELU decreased equivalent carbon footprint by about 17%. We expect that eco2AI could help the ML community to pace to Green and Sustainable AI within the presented concept of AI-based GHG sequestrating cycle.

https://github.com/sb-ai-lab/Eco2AI

• International

Yes. Every person anywhere in the world can use the project with help of GitHub and PyPI

This project motivates optimized patterns of power consumption and carbon footprint of data centers promoting sustainable economy.

Reduction of carbon footprints capable of slowing down Global Warming. It also encourages the research community to utilize AI more consciously.