search

starry-sky6688 / MADDPG

Pytorch implementation of the MARL algorithm, MADDPG, which correspondings to the paper "Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments".
537 stars 83 forks source link

其他两个球学不到正确的策略 #42

Closed Rhoslynnn closed 7 months ago

Rhoslynnn commented 7 months ago

我将ANN替换成了SNN，接口都和您的代码保持一致，但是回报率一直稳定在600左右，最后发现是只有一个球可以学到正确的策略，其他两个球都在原地打转，可以问一下是什么原因导致的吗

starry-sky6688 commented 7 months ago

既然只换了神经网络，那多半是网络的问题了