Maximum entropy GFlowNets with soft Q-learning