Reinforcement Learning on the Discrete Composition Channel of a Crystal Generator: Validated Gains and Reward Hacking
My rating
The paper targets inverse materials design, where a generative model must optimize crystalline materials for specific properties rather than merely reproduce a structural database. It uses group-relative policy optimization to align a generative model based on stochastic interpolants and discrete flow matching. The study also reports validated gains and investigates reward hacking in the reinforcement-learning setup.
Tags: