InvalidArgumentError

[longqianlanqiu]

尝试使用paddle的DQN算法训练gym中的环境时，出现了如下报错： InvalidArgumentError: After flatten the input tensor X and Y to 2-D dimensions matrix X1 and Y1, the matrix X1's width must be equal with matrix Y1's height. But received X's shape = [1, 210, 160, 3], X1's shape = [1, 100800], X1's width = 100800; Y's shape = [210, 128], Y1's shape = [210, 128], Y1's height = 210. [Hint: Expected x_mat_dims[1] == y_mat_dims[0], but received x_mat_dims[1]:100800 != y_mat_dims[0]:210.] at (D:\1.8.5\paddle\paddle\fluid\operators\mul_op.cc:83) [operator < mul > error] 请问是哪里出错？ 训练的环境是：SpaceInvaders-v0

[TomorrowIsAnOtherDay]

你使用的是PARL仓库中的代码吗？还是用paddle自己写的代码呢？

[longqianlanqiu]

是使用的仓库的代码

[TomorrowIsAnOtherDay]

了解了，还请把你的paddle、parl版本发一下。 我们这边本地先复现问题再提供解决方案哈。

[longqianlanqiu]

paddle 版本是1.8.5CPU parl版本是1.4.3

[TomorrowIsAnOtherDay]

收到：）感谢反馈

[TomorrowIsAnOtherDay]

hello，我发现新版本的atari-py 以及gym向后兼容不友好。 建议安装如下版本的：

pip install atari-py==0.1.15
pip install gym=0.12.0