We are two data scientists, Chrisanna Cornish and Danielle Dequin working together to evaluate the carbon footprint of data science competitions. Under the guidance of research professional Veronika Chelpygina we are evaluating the energy consumption of kaggle competitions, starting with SIIM-FISABIO-RSNA COVID-19 Detection.
We measure the cost of training two models on seven datasets. The following are the links to the original repositories for the models and their respective datasets:
Within our organization you will find forks of the above repositories with any necessary changes we made to run the code locally and add CarbonTracker to estimate the carbon footprint of training. The links to those forked repositories are: