{
  "name": "chembl36_selfies",
  "dataset_name": "lukaskim/ChEMBL-36",
  "dataset_config": "molecules",
  "source_split": "train",
  "representation": "SELFIES",
  "smiles_column": "canonical_smiles",
  "selfies_column": "selfies",
  "canonical_smiles_column": "smiles_canonical_clean",
  "config": {
    "dataset_name": "lukaskim/ChEMBL-36",
    "dataset_config": "molecules",
    "split": "train",
    "smiles_column": "canonical_smiles",
    "output_dir": "data/pretrain/chembl36_selfies",
    "seed": 13,
    "valid_fraction": 0.01,
    "test_fraction": 0.0,
    "max_rows": null,
    "dedupe_column": "standard_inchi_key",
    "min_heavy_atoms": 3,
    "max_heavy_atoms": 100,
    "max_mw": 1000.0
  },
  "source_row_count": 2854815,
  "preparation_stats": {
    "input_rows": 2854815,
    "dedupe_column": "standard_inchi_key",
    "rows_after_dedupe": 2854815,
    "rows_valid_after_conversion": 2854762,
    "rows_after_filters": 2414542,
    "dropped_invalid_or_filtered": 440273,
    "sanitize_error_counts": {
      "failed_basic_filters": 440220,
      "rdkit_parse_failed": 15,
      "selfies_failed:EncoderError": 38,
      "valid": 2414542
    }
  },
  "row_counts": {
    "prepared_total": 2414542,
    "train": 2390314,
    "valid": 24228
  },
  "columns": [
    "chembl_id",
    "canonical_smiles",
    "standard_inchi",
    "standard_inchi_key",
    "molecule_type",
    "max_phase",
    "first_approval",
    "oral",
    "prodrug",
    "natural_product",
    "black_box_warning",
    "withdrawn_flag",
    "therapeutic_flag",
    "mw_freebase",
    "alogp",
    "hba",
    "hbd",
    "psa",
    "rtb",
    "aromatic_rings",
    "heavy_atoms",
    "qed_weighted",
    "num_ro5_violations",
    "smiles_canonical_clean",
    "selfies",
    "is_valid",
    "sanitize_error",
    "split_key"
  ],
  "split_overlap": {
    "train_valid": {
      "n_overlap": 1,
      "examples": [
        "CC(C)(NS(=O)(=O)c1ccccc1)C(=O)NC1C2CC3CC1CC(C(N)=O)(C3)C2"
      ]
    }
  },
  "split_policy": {
    "method": "deterministic_hash",
    "key_column": "split_key",
    "seed": 13,
    "valid_fraction": 0.01,
    "test_fraction": 0.0,
    "has_test": false
  },
  "versions": {
    "datasets": "4.8.5",
    "rdkit": "2026.03.1",
    "selfies": "2.1.1",
    "pandas": "3.0.3",
    "pyarrow": "24.0.0"
  },
  "created_at_utc": "2026-05-15T11:54:13.343104+00:00",
  "creation_command": "python -m modernmolbert.data.prepare_chembl36_selfies"
}