AI RESEARCH
C-MORAL: Controllable Multi-Objective Molecular Optimization with Reinforcement Alignment for LLMs
arXiv CS.LG
•
ArXi:2604.23061v1 Announce Type: new Large language models (LLMs) show promise for molecular optimization, but aligning them with selective and competing drug-design constraints remains challenging. We propose C-Moral, a reinforcement learning post-