1
Controllable Molecular Generative Foundation Models
提出首个统一框架，用强化学习后训练攻克分子图生成的可控性难题，解决原子级动作空间和化学无效中间态问题
arXiv:2605.15354v1 Announce Type: new Abstract: Despite the success of foundation models in language and vision, molecular graph generation still lack…