docs/memo_assets/split_source_overlap_summary.json

{
  "unique_sources": {
    "train": 20,
    "val": 20,
    "test": 20
  },
  "pairwise_overlap": {
    "train_val": 20,
    "train_test": 20,
    "val_train": 20,
    "val_test": 20,
    "test_train": 20,
    "test_val": 20
  }
}