CRNDiff: Count-Native Diffusion Framework via Chemical Reaction Networks
CRNDiff, a structured framework that combines count-space diffusion with inference-time conditioning on rare subpopulations, and tilted Feynman--Kac steering, a method for sampling target subpopulations from a frozen generator without retraining are introduced.
Yu-Shan Qiu, Praful Gagrani, Tetsuya J. Kobayashi
· 0 citations