Estimate and track the environmental footprint of GenAI models at inference

EcoLogits, an open source project under the CodeCarbon non-profit, offers a suite of tools for estimating the environmental footprint of generative AI models at inference, built on life-cycle assessment principles. The suite includes a Python library that tracks impacts of AI requests made through official client SDKs, a JavaScript and TypeScript library published as @ecologits/ecologits.js on npm, an API service for embedding the impact assessment methodology into applications, and CodeCarbon for tracking energy use and CO₂ emissions of code on local hardware or cloud VMs. Data For Good and Boavizta supported the project's development with tools, best practices and environmental impact assessment expertise.

Project EcoLogits is a suite of open source tools for estimating the environmental footprint of generative AI models at inference . Based on life-cycle assessment principles , the project raises awareness about the direct environmental impacts of AI while empowering developers and organizations to build more sustainable AI-powered applications . - Open & transparent – Code, methodology, and data are openly accessible. - Ease of use – Emphasizes on seamless integration and user experience. - Community-driven – Continuously built and improved collaboratively. Tools EcoLogits Calculator EcoLogits A Python library to track environmental impacts of AI requests made through official client SDKs. Learn more ../get started EcoLogits.js A JavaScript and TypeScript library to estimate the environmental impacts of generative AI. Learn more https://www.npmjs.com/package/@ecologits/ecologits.js EcoLogits API An API service enabling integrations of our impact assessment methodology into your applications. Learn more https://api.ecologits.ai/docs CodeCarbon Local compute Track the energy use and CO₂ emissions of code running on your own hardware or cloud VMs. Learn more https://docs.codecarbon.io/latest/ Contributors People who contribute significantly to the project. Sponsors Organizations who make this project possible. Acknowledgements EcoLogits is part of the CodeCarbon https://codecarbon.io/ non-profit. Together aim to build open and robust tools to assess and reduce the environmental footprint of AI. We extend our gratitude to Data For Good https://dataforgood.fr/ and Boavizta https://boavizta.org/en for supporting the development of this project. Their contributions of tools, best practices, and expertise in environmental impact assessment have been invaluable.