1 paper
Yunkai Yang, Yudong Zhang, Xinying Chen +5
Reinforcement Learning (RL) has proven effective in aligning flow-based generative models with human preferences. Recently, Flow-GRPO has emerged as an efficient critic-free paradi…