Tag
PerturbCellRL introduces a reinforcement learning framework that post-trains a pretrained single-cell transcriptomic generator using cell-level verifiers as rewards, improving biological consistency of perturbation predictions beyond distributional matching.