1 paper across 1 session
We propose a flexible two-stage MOF generation framework that combines a SMILES-based autoregressive model with a torsion-aware normalizing flow.