Run to the tenth episode and report an error

SukerZ / MADDPG-on-PyTorch

在PyTorch上重构multi-agent deep deterministic policy gradient(MADDPG)，将https://github.com/xuemei-ye/maddpg-mpe 修改到自己电脑上可运行。因为本人笔记本没有CUDA，实验速度太慢，不能保证算法被正确实现。

13 stars 3 forks source link

Run to the tenth episode and report an error #1

Open 13276019110 opened 3 years ago

13276019110 commented 3 years ago

C:\Users\builder\AppData\Local\Temp\pip-req-build-r5jc5joa\aten\src\ATen/native/IndexingUtils.h:20: UserWarning: indexing with dtype torch.uint8 is now deprecated, please use a dtype torch.bool instead. Warning: indexing with dtype torch.uint8 is now deprecated, please use a dtype torch.bool instead. (expandTensors

How should I improve the code to avoid this problem

mjy618 commented 2 years ago

At this time, the memory is not full, you should set the memory to be full before starting training.

b762927 commented 2 years ago

Why we train this project at the 10th episode, this project start work very slowly and almost stop...

SukerZ commented 1 month ago

Thanks you all. Because I wrote this project only for learning the multi-agent reinforcement learning algorithms，I have not run to the tenth episode, I don't know what went wrong, sorry.