artifacts/summary/FROZEN_CONFIG.json

{
  "analysis_id": "reviewer_requested_reanalysis_v4",
  "input_csv": "train.csv",
  "output_root": "revision/artifacts/reviewer_requested_reanalysis_v4",
  "outer_seeds": [
    123456,
    123457,
    123458
  ],
  "inner_folds": 6,
  "split_strategies": {
    "canonical_grouped": {
      "group_column": "structure_group",
      "outer_folds": 10,
      "balance_group_sizes": false
    },
    "scaffold_aware": {
      "group_column": "scaffold_component_group",
      "outer_folds": 5,
      "balance_group_sizes": true
    }
  },
  "descriptor_features": [
    "LogP",
    "BertzCT",
    "MolMR",
    "NumAromaticRings",
    "HeavyAtomCount"
  ],
  "fingerprint": {
    "radius": 2,
    "n_bits": 2048,
    "use_chirality": true
  },
  "classical_baselines": {
    "n_estimators": 500,
    "max_depth": null,
    "min_samples_leaf": 2,
    "max_features": 1.0,
    "n_jobs": -1
  },
  "bootstrap": {
    "n_resamples": 2000,
    "confidence": 0.95
  },
  "prospective_domain": {
    "similarity_thresholds": [
      0.3,
      0.4,
      0.5
    ]
  },
  "neural": {
    "base_models": [
      "gat",
      "gcn",
      "fpnn"
    ],
    "device": "auto",
    "meta_learner": {
      "n_estimators": 800,
      "max_depth": 12,
      "min_samples_split": 5,
      "min_samples_leaf": 3,
      "max_features": 0.8,
      "bootstrap": true,
      "max_samples": 0.85,
      "n_jobs": -1
    }
  },
  "outer_folds": 10
}