GFlowNet Training by Policy Gradients https://arxiv.org/abs/2408.05885 Haven't read this yet, but seems important.