Off-policy reinforcement learning (RL) tries to learn the value of a target policy while data comes from a different behavior policy. This often requires tra…
机构:DeepMind
来源:arXiv 2611.13275 | AI4Papers 论文推荐平台