This repository contains code and models used to compete in the competition: Deep Learning for Climate Emulation
Our team HK(N+1) achieved second place, with a private leaderboard score of 0.65.
Repository organization:
(●'◡'●)
├─ .gitignore
├─ README.md
├─ climate_prediction
│ ├─ __init__.py
│ ├─ data_loader.py // data module (both regular and decadal)
│ ├─ loss.py // custom loss
│ ├─ models
│ │ ├─ CNN_LSTM.py
│ │ ├─ CNN_MLP.py
│ │ ├─ ClimaX_Transformer.py
│ │ └─ ConvLSTM.py
│ ├─ train.py // lightning module
│ └─ util.py // utilities, visualizations, etc.
├─ notebooks
│ ├─ PR-CNN-MLP.ipynb
│ └─ TAS_ClimaX.ipynb
└─ requirements.txt