AI RESEARCH
Moment Matching Q-Learning
arXiv CS.LG
•
ArXi:2605.29033v1 Announce Type: new Score-based and flow-based generative models exhibit remarkable expressive capacity in capturing complex distributions, and have been extensively deployed in tasks ranging from image generation to reinforcement learning. Nevertheless, these models suffer from prolonged inference latency, which imposes a significant computational bottleneck in RL with iterative sampling.