CUDA out of memory

[AragornKris]

Hello, I use a RTX2080ti to train the dataset you provided and do not change any code, but an "CUDA out of memory" error occurred in the program. Could you give some help? Thank you !

[PennyPeng369]

I met the same issue. I think the GPU memory required may be >= 12G. I havn't found any workable method to solve it. Any thought of solution would be appreciated. Thanks.

[AragornKris]

I met the same issue. I think the GPU memory required may be >= 12G. I havn't found any workable method to solve it. Any thought of solution would be appreciated. Thanks.

I delete part of the parallel calculation code(in solver.py from line125 to line134), it can work.

[AloneGu]

Replace all 64 with 32 of Generator.__init__ ( in net.py ), the max gpu memory usage is around 7500MiB, it goes well with my GTX 1070

[qsong2001]

export CUDA_VISIBLE_DEVICES=0 before running python main.py It seems parallel calculation needs the GPU memory >=12G.

[AragornKris]

这是来自QQ邮箱本人忙碌的回复   您好，我最近很忙，无法亲自回复您的邮件。我忙完后，会尽快给您回复。

[wtjiang98]

Replace all 64 with 32 of Generator.__init__ ( in net.py ), the max gpu memory usage is around 7500MiB, it goes well with my GTX 1070

That's a good idea. Just reduce the dimension of the bottlenecks.