Training data for arXiv:2401.04070

The pickle file contains the training data for arXiv:2401.04070

To access it, use a python script with the following

```python

import pickle

with open("train_val_test.p", "rb") as f:

data = pickle.load(f)

```

This will give a huge dictionary with the following structure

{"train": {"structures": [], "energies": [], "forces": [], "stresses":[], "id": []}, "validation": xxx, "test": ...