ai4climate

AI4Climate: Collection of Machine Learning Tasks and Datasets for Tackling Climate Change

Overview

1. Getting started
2. Available datasets
3. Contributions

Getting started

All datasets are provided on Hugging Face Hub and ready to be downloaded and parsed into our standardized data format with training, validation and testing splits using our ai4climate Python package. Install package:

pip install ai4climate

For example, load the "train_small_test_medium" task from the "OPFData" dataset:

from ai4climate import load

dataset = load(
    task_name='OPFData', 
    subtask_name='train_small_test_medium',
    root_path='~/AI4Climate/'
)

Run main experiments as specified in config.yml with:

python src/main.py

List of available datasets

1. OPFData
2. PowerGraph
3. SolarCube
4. BuildingElectricity
5. WindFarm