Closed luo-li-ba-suo closed 3 years ago
Sorry, I don't speak Chinese.
I have found the reason why it comes to not enough memory: the scores and the losses in DQNAgent should be manually constrainted to a capicity that is not too large.Thank you
Also you can reduce the mini batch size or try it on Colab. Thanks for sharing the reason of the issue. I will close this issue.
是缓存器的问题吗