Representation-Guided Discrete Molecular Graph Retrosynthesis

ArXi:2605.24428v1 Announce Type: new Stochastic process-based molecular graph generators have become the state of the art for template-free single-step retrosynthesis. However, these models are typically trained only on product-reactant pairs, thereby acquiring chemistry-relevant representations in an indirect and implicit manner. Meanwhile, recent advances in computer vision nstrate that offering representation guidance to a generator can effectively distill semantics from pretrained encoders into DiTs, substantially improving both convergence and generation quality.