from pino.upload_data import create_molecule_disjoint_split

for ratio in [0.4, 0.45, 0.5, 0.55, 0.6]:
    dd = create_molecule_disjoint_split('data/synthetic_dataset_v2.jsonl', train_ratio=ratio, seed=42)
    print("ratio=%.2f train=%d val=%d" % (ratio, len(dd['train']), len(dd['validation'])))