AI RESEARCH

Efficient Adjoint Matching for Fine-tuning Diffusion Models

arXiv CS.LG

ArXi:2605.11480v1 Announce Type: new Reward fine-tuning has become a common approach for aligning pretrained diffusion and flow models with human preferences in text-to-image generation. Among reward-gradient-based methods, Adjoint Matching (AM) provides a principled formulation by casting reward fine-tuning as a stochastic optimal control (SOC) problem.