CodeCarbon

Python tool to track and reduce CO2 emissions from ML training

Code & DevelopmentFree

CodeCarbon is an open-source Python package that estimates the carbon dioxide emissions produced by running machine learning and AI code. It hooks into your training scripts to track compute usage, maps that to energy consumption based on your hardware, and then calculates CO2 equivalent based on your electricity grid's carbon intensity. CodeCarbon helps ML teams understand and report the environmental cost of their AI work and find more efficient alternatives.

Key Features

• ✓Emissions tracking
• ✓Energy monitoring
• ✓Carbon intensity by region
• ✓Experiment logging
• ✓Dashboard visualization
• ✓Open-source