{ "deployment_refit": true, "in_sample_diagnostics_ONLY_not_validation": { "cls_scaffold": 1.0, "cls_random": 1.0, "reg_scaffold": 9.239338538866721e-11, "reg_random": 9.239338538866721e-11 }, "hard_gate_report": { "per_set": { "herg70": { "unique_keys": 462, "connectivity_in_cls_file": 462, "in_cls_training_matrix": 462, "full_ikey_in_cls_file": 452, "connectivity_in_reg_file": 462, "in_reg_training_matrix": 462, "full_ikey_in_reg_file": 452 }, "herg60": { "unique_keys": 249, "connectivity_in_cls_file": 249, "in_cls_training_matrix": 249, "full_ikey_in_cls_file": 245, "connectivity_in_reg_file": 249, "in_reg_training_matrix": 249, "full_ikey_in_reg_file": 245 }, "new_compounds": { "unique_keys": 967, "connectivity_in_cls_file": 967, "in_cls_training_matrix": 967, "full_ikey_in_cls_file": 967, "connectivity_in_reg_file": 967, "in_reg_training_matrix": 967, "full_ikey_in_reg_file": 967 }, "fda": { "unique_keys": 36, "connectivity_in_cls_file": 36, "in_cls_training_matrix": 36, "full_ikey_in_cls_file": 34, "connectivity_in_reg_file": 36, "in_reg_training_matrix": 36, "full_ikey_in_reg_file": 34 }, "regfail": { "unique_keys": 7, "connectivity_in_cls_file": 7, "in_cls_training_matrix": 7, "full_ikey_in_cls_file": 6, "connectivity_in_reg_file": 7, "in_reg_training_matrix": 7, "full_ikey_in_reg_file": 6 }, "test1": { "unique_keys": 43, "connectivity_in_cls_file": 43, "in_cls_training_matrix": 43, "full_ikey_in_cls_file": 41, "connectivity_in_reg_file": 17, "in_reg_training_matrix": 17, "full_ikey_in_reg_file": 15 } }, "cls_alldata_n": 14441, "inchikey_compute_failures_cls": 0, "connectivity_vs_compound_key_mismatches_cls": 0, "store_missing_cls": 0, "bundle_n_train_cls": 14441, "reg_alldata_n": 10370, "inchikey_compute_failures_reg": 0, "connectivity_vs_compound_key_mismatches_reg": 0, "store_missing_reg": 0, "bundle_n_train_reg": 10370, "pass": true }, "honest_locked_benchmark_scores": { "source": "Locked-pool Phase-5 refits (train+val only; every external/regulatory set locked out), results/phase5/PHASE5_RESULTS.md \u2014 these are the honest benchmark scores of this architecture. The DEPLOYMENT refits in this repo absorbed those sets and are never scored on them.", "classification_scaffold_refit": { "hERG-70": { "ACC": 0.76, "MCC": 0.543 }, "new_compounds_MCC": 0.149, "FDA_errors": "4/36", "REGFAIL_Tier1": "4/5" }, "classification_random_refit": { "hERG-70": { "ACC": 0.74, "MCC": 0.504 }, "new_compounds_MCC": 0.154, "FDA_errors": "5/36", "REGFAIL_Tier1": "4/5" }, "regression_scaffold_refit": { "hERG-70_RMSE": 0.838 }, "regression_random_refit": { "hERG-70_RMSE": 0.829 }, "note": "Best SINGLE (non-fusion) model on hERG-70 in the programme (MCC 0.543); the best overall classifier is Fusion G (0.548, XGBoost + CardioGenAI), which is not a deployment model." }, "note": "In-sample diagnostics are printed for sanity only (ExtraTrees in-sample accuracy is 1.0000 and in-sample regression RMSE ~0.0000 by construction of the algorithm \u2014 fully grown trees memorize the training rows; they are NOT performance estimates). These refits absorbed every external/regulatory set, so no honest external validation of them exists; the honest scores of this architecture are the locked-pool Phase-5 refits listed under honest_locked_benchmark_scores." }