Nature | Biotechnology Follow Deep learning perturbation models can outperform baselines on calibrated metrics Calibration-aware evaluation of perturbation models shows that deep learning models can outperform baselines. https://www.nature.com/articles/s41587-026-03307-w nature.com
