1 paper
Zheming Fu, Ruizhe He, Wei Shang +5
Aligning generative flow models on continuous spaces via online reinforcement learning is constrained by intractable trajectory likelihoods. Existing density-approximated policy gr…