Smoothing Dark Areas in Molecular Latent Diffusion
A topology-optimized VAE that creates smoother, more chemically valid latent spaces for 3D molecular diffusion.
Research archive
Research on generative models, spanning LLM post-training, diffusion models for drug discovery, and foundation models for RNA, proteins, and antibodies.
A topology-optimized VAE that creates smoother, more chemically valid latent spaces for 3D molecular diffusion.
A multimodal benchmark for evaluating how well language models understand chemical reactions in scientific literature.
A new trajectory-balance objective and diversity-promoting replay strategy for stable GFlowNet training.
A benchmark for evaluating whether molecular representations preserve geometry, chirality, and conformational energy.
A spatial representation method for molecular chirality classification using learnable queries.
An expert-curated, open database of chiral ligands and catalysts for asymmetric synthesis and data-driven research.
Classifier-guided diffusion for controllable atropisomer generation in data-scarce regimes.
A substrate-aware descriptor that improves reaction-performance prediction while preserving interpretability.
Large-scale RNA pretraining that captures evolutionary and structural information for downstream prediction.
A review of synergistic strategies for stereocontrol in palladium-catalyzed asymmetric allylic substitution.
A review of bimetallic catalytic systems for stereodivergent synthesis across major reaction classes.