| { | |
| "exclusions": [ | |
| { | |
| "feature": "aroma_chemicals.openpom_embedding", | |
| "reason": "no provenance field or real model asset; implementation generates 128-bit Morgan plus 10 physicochemical descriptors" | |
| } | |
| ], | |
| "frozen_at": "2026-07-17T05:12:56.027886+00:00", | |
| "frozen_method": { | |
| "folds": "five deterministic grouped folds; CAS and formula families may not cross folds", | |
| "metrics": "Spearman rho/MAE for continuous; macro-F1/AUROC for categorical", | |
| "missing_values": "exclude a row from all representations if any preregistered input is missing", | |
| "probe": "standardized ridge regression for continuous targets; logistic ridge for categorical targets", | |
| "row_policy": "complete-case intersection per task, fixed before representation access", | |
| "selection": "regularization selected by nested CV using training folds only", | |
| "uncertainty": "paired fold deltas with percentile bootstrap 95% CI" | |
| }, | |
| "representations": { | |
| "morgan": { | |
| "available": true, | |
| "definition": "RDKit Morgan radius=2" | |
| }, | |
| "pom_plus_physics": { | |
| "available": false, | |
| "definition": "real_pom concatenated with preregistered physical features" | |
| }, | |
| "real_pom": { | |
| "available": false, | |
| "definition": "independently loaded pretrained POM model output" | |
| } | |
| }, | |
| "results": null, | |
| "schema_version": 1, | |
| "status": "blocked_missing_real_pom", | |
| "tasks": [ | |
| "descriptor", | |
| "odor_threshold", | |
| "substantivity", | |
| "substitution" | |
| ] | |
| } | |