← All datasets

Molecular properties

Aqueous Solubility (AqSolDB)

AqSolDB collects experimental measurements of how readily compounds dissolve in water. The benchmark predicts logS from molecular fingerprints, with logS defined as the base-10 logarithm of solubility in moles per litre.

RegressionLicence: CC BY 4.0 via TDC; underlying source terms require verification before re-hostingTDC-AqSolDBTarget: Ytdc ↗

Mean cross-validation scores for this dataset. Each score comes from the requested feature- and sample-cap cell, and failed fits use the constant predictor.

At this operating point

Model comparison

p=10,000, n=100 · Lower RMSE is better

Random Forest is marked with a diamond. Models are ordered by the selected metric.

Across sample budgets

Learning curves

All metric values at this operating point
ModelRMSEMAER²
TabPFN 32.0921.6410.1284
Random Forest2.0981.6260.1422
Extra Trees2.0991.6280.1419
TabM2.1061.6730.1247
AutoGluon2.1071.6520.1294
TabICL v22.1071.6520.129
RealTabPFN 2.52.1191.6560.1172
CatBoost2.131.6620.1174
RealTabPFN v22.1341.6940.1132
MLP2.1481.70.07983
Logistic Regression2.1681.6890.08293
MITRA2.1721.7030.05265
XGBoost2.2031.7250.03555
RealMLP2.2631.7530.002899
TabFM2.2641.788-0.07479
TabDPT †2.2721.760.01669
KNN2.2841.807-0.009969
LightGBM2.2841.83-0.02366