Molecular properties

Glucocorticoid Receptor Bioactivity

This dataset pairs chemical structures with measured bioactivity against the human glucocorticoid receptor. The benchmark represents each molecule as a fingerprint and predicts its median pChEMBL value, which summarizes potency or affinity on a logarithmic scale.

RegressionLicence: CC BY-SA 3.0CHEMBL2034Target: pchembl_valuechembl ↗

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

Models

All metric values at this operating point

Model	RMSE	MAE	R²
LimiX2	0.9272	0.7144	0.264
TabPFN 3.5	0.9346	0.7228	0.2541
RealTabPFN 2.5	0.9388	0.7312	0.2478
AutoGluon	0.9432	0.7269	0.2409
Random Forest	0.9455	0.7279	0.237
Extra Trees	0.9474	0.7297	0.234
Random Forest (tuned)	0.9497	0.7353	0.2301
RealTabPFN v2	0.9507	0.7428	0.2287
Kumo Tabular (M)	0.9525	0.7241	0.2258
GeneICL	0.9527	0.7422	0.2256
Logistic Regression	0.9569	0.7391	0.2185
Logistic Regression (tuned)	0.9571	0.7388	0.2184
TabFM	0.9623	0.7321	0.2091
MLP	0.9638	0.7466	0.2081
TabM	0.9644	0.7595	0.2054
MITRA	0.9657	0.7567	0.2031
TabDPT	0.9749	0.7419	0.189
XGBoost	0.9751	0.7582	0.1879
RealMLP	0.9758	0.764	0.1866
MLP (tuned)	0.9781	0.756	0.1829
XGBoost (tuned)	0.9912	0.7737	0.1609
Extra Trees (tuned)	1.021	0.781	0.09685
Causilo	1.031	0.7916	0.09211