Deep learning generative models hold significant promise for exploring the configuration space of crystalline materials. Here, the authors present CrystalFlow, a flow-based generative model for ...