The campaign.tar.gz archive contains several subfolders:

- `./datasets`: The design space datasets used here, in the OPTIMADE JSONL format.
- `./epochs`: The results of each active learning epoch, i.e., calculations of the selected materials
- `./features`: The features computed for the design space datasets
- `./models`: Checkpoints for the MODNet models trained at each epoch: these can be loaded via the methods in the `modnet` Python library.
` `./results`: CIF files for each of the structures calculated here, plus a `results.csv` providing the computed properties for each structure, with provided OPTIMADE IDs.