PublicDatasetModelMetrics

  • Human-readable name of the selected algorithm.

  • Name of the primary performance metric, such as r2_cv_mean or mcc_cv_mean.

  • Primary cross-validation or test-set performance value. Interpret it using correlation_validation_name.

  • Name of the secondary performance metric, such as mae_cv_mean or roc_auc_cv_mean.

  • Secondary cross-validation or test-set performance value. Interpret it using error_validation_name.

  • Human-readable name of the selected molecular representation, including the foundation model name when one was selected.

  • Whether the model was trained on log-transformed targets.

  • R-squared on user-provided external test data, if any.