Machine Learning For Graphs Project

Setup:

1. cd graphcast
2. pip install -e .
3. pip install torch

The files:

• updated_model.py – For debugging the updated GraphCast model
• updated_model.ipynb – Similar to the Python file, but also has test MLP and saves the acquired data
• getting_data.ipynb – Loading in example data, model weights, and streaming ERA5 dataset
• full.py and half.py – For running the experiments and generating data. Use respective .job files with Snellius
• plots.ipynb – Generating plots and analysis

Modified GraphCast files:

• deep_typed_graph_net.py
• graphcast.py
• casting.py
• normalization.py
• autoregressive.py
• rollout.py