Tabular learning × biomedicine

A benchmark for high-dimensional biomedical tables.

TabBench-Bio compares classical models, neural networks, and tabular foundation models across transcriptomic, metagenomic, molecular, sequence-derived, and genomic-prediction datasets on a controlled feature-by-sample grid.

Read on arXiv View on GitHub Request dataset / model

—biomedical datasets

—model configurations

—evaluation points per model

5×cross-validation

Interactive results

Model rankings depend on the operating point.

Leaderboard

Bradley–Terry Elo

Metric Dataset modality Feature cap Training samples

Random Forest is anchored at 1,000. Error bars are target-bootstrap 95% intervals.

Budget response

Performance across sample budgets

Dataset modality Feature cap Models UncertaintyShow 95% CI

Efficiency

Performance vs. fitting cost

Metric Dataset modality Feature cap Training samples

Efficiency

Performance vs. prediction cost

Metric Dataset modality Feature cap Training samples

Stability

Rank stability across operating points

Reference cell

Coverage

Datasets by modality

Snapshot scope. The live counter covers the configured benchmark run; performance charts remain pinned to the last status-complete aggregation until the cells finish.

Strict primary analysis. By default, performance charts and rankings use results from the requested nominal cell. Attempted failures follow the benchmark's explicit failure policy rather than borrowing performance from a smaller sample size. The adaptive sensitivity view follows the same model, fold, dataset, and feature cap through smaller configured sample cells after a training out-of-memory failure; it is not interpreted as performance at the requested larger sample size. Prediction-only OOMs instead reduce the query batch without changing the trained model or test set.

A benchmark for high-dimensional biomedical tables.

Model rankings depend on the operating point.

Bradley–Terry Elo

Performance across sample budgets

Performance vs. fitting cost

Performance vs. prediction cost

Rank stability across operating points

Datasets by modality

Explore the benchmark in detail.