Sberbank is the largest bank in Russia, Central and Eastern Europe, one of the leading financial institutions globally.
The library is intended to monitor the carbon footprint of the AI model training process. It determines available computing resources and calculates how much energy was consumed in the training process, and equivalent CO₂ emissions, taking regional standards into account.
International project
E
2022
-
Growing energy consumption on AI model training and corresponding growth in the carbon footprint. Absence of tools allowing to measure or manage this issue.
The goal is to create a library that would track energy consumption and carbon footprint, thereby encouraging the DS community to develop computer-optimal machine-learning architectures. Objectives: develop a carbon footprint calculation methodology, create the library, publish the library on an open-source basis.
Companies and teams that train AI models.
-
24+ thousand downloads
-
-
