split_validation.json

{
  "passed": true,
  "pcqm_rows": 500000,
  "pcqm_splits": {
    "train": 489511,
    "validation": 10403,
    "cross_objective_holdout": 69,
    "benchmark_overlap": 16,
    "identity_quarantine": 1
  },
  "cosmo_rows": 360259,
  "cosmo_splits": {
    "train": 352834,
    "validation": 7082,
    "new_holdout": 168,
    "cross_objective_holdout": 92,
    "benchmark_overlap": 83
  },
  "benchmark_test_solute_groups": 126,
  "no_pretraining_train_benchmark_overlap": true,
  "no_cross_objective_train_validation_group_overlap": true,
  "pcqm_composition_validated_all_rows": true,
  "pcqm_invalid_identity_rows": 1,
  "all_train_compositions_and_charges_validated": true,
  "geometry": "Original PCQM SDF coordinates and original COSMO PM6 coordinates; no regeneration",
  "initial_rem3di_checkpoint": null,
  "combisolv_training": false
}