Designed by the University of Copenhagen as part of the enRichMyData toolbox, Carbontracker is a cutting-edge tool designed to seamlessly measure and predict the carbon footprint of machine learning models. Whether you are a developer, researcher, or data scientist, it provides a clear picture of the environmental impact of your training processes.
Key Features
- Real-Time Monitoring: Tracks power consumption and carbon emissions during model training.
- Localized Accuracy: Uses real-time data on local energy carbon intensity for precise measurements.
- Broad Hardware Support: Compatible with Intel/AMD CPUs, NVIDIA GPUs, and Apple silicon.
- Informed Decisions: Estimates total emissions after training to help you optimize your workflows sustainably.
Effortless Integration
Carbontracker is designed to fit right into your workflow.
- Command-Line Interface (CLI): For quick and easy usage.
- Python Bindings: Perfect for seamless integration into machine learning pipelines.
Transparency and Accessibility
As an open-source tool available on GitHub under the MIT license, Carbontracker promotes accessibility and transparency. The website, carbontracker.info, provides:
- A detailed roadmap.
- Comprehensive documentation.
- Practical examples to help users get started.
Why It Matters
By leveraging real-time, localized carbon intensity data, Carbontracker empowers users to take meaningful steps towards sustainable AI. It’s more than just a tracker; it’s a way to reconcile innovation with responsibility.
Get Started Today
Ready to make your machine learning workflows more eco-friendly? Check out Carbontracker and join the movement toward a greener AI future.
GitHub: Carbontracker Learn More: carbontracker.info