#Adjoint Sampling

Meta AI released Adjoint Sampling, a new scalable algorithm that trains generative models using only scalar reward signals, bypassing the need for large datasets and enabling advances in molecular modeling.