RuntimeError : cuda runtime error(11) : invalid argument at /pytorch/aten/src/THC/THCgeneral.cpp:405

[Realdr4g0n]

---

Hi. I've seen your research very interesting.
I am trying to train a model on github that I am interested in.

My gpu enviroment is:

After doing dataset generatation as described in git

sudo CUDA_VISIBLE_DEVICES=0 python3 train.py --dataroot ./datasets/jeans2skirt_ccp --model insta_gan --name jeans2skirt_ccp_instagan --loadSizeH 330 --loadSizeW 220 --fineSizeH 300 --fineSizeW 200 --niter 400 --niter_decay 200

When I execute the above, I get the following error.

I know it is a Cuda version issue, so if you run it in various versions, it will appear as well.
Thank you for your understanding. What I wrote in Korean

[sangwoomo]

Hello, I comment in English as it may be helpful for others with same issue.
First, can you try to use --gpu_ids 0 instead of sudo CUDA_VISIBLE_DEVICES=0?

[Realdr4g0n]

Thanks for your comment.

Yes I've tried --gpu_ids 0. But same error occured

[Realdr4g0n]

Can I know about your development environment about cuda and cudnn version?

[sangwoomo]

I used CUDA 9.0, CUDNN 7.0, and PyTorch 0.4.
I thinks it's a PyTorch bug. Does other codes work in your environment?

It seems that the latest PyTorch is not compatible with CUDA < 10.0.
https://discuss.pytorch.org/t/cuda-runtime-error-11/30080/10

[Realdr4g0n]

Thanks for your feedback
As I said above, I am using cuda 10.0 version, this error is caused by PyTorch version.

I've use torch version 1.0.0. But I uninstall this package and now I've change to version 0.4
so it's work!

Thanks a lot

[AliaChen]

Hi, I have the same erro about cuda run time erro(11), I changed the cuda version as 9.0,but it does not work?

[sangwoomo]

Hi, did you use PyTorch 0.4?

[AliaChen]

Hi, did you use PyTorch 0.4?

yes,I used Pytorch0.4.1 ,and GeForce RTX2080ti.

[sangwoomo]

Hi, it seems that there is an incompatibility of PyTorch with some environments.
I'm sorry that I'm not maintaining this code base, and it'd be really appreciate if you find some solution and share it (may be from the official PyTorch github)!

[AliaChen]

Hi, it seems that there is an incompatibility of PyTorch with some environments.
I'm sorry that I'm not maintaining this code base, and it'd be really appreciate if you find some solution and share it (may be from the official PyTorch github)!

Gratefully, but I still have not solved this issue. If it is solved, I will share it with you here.

[meincv]

Is anybody find a solution?

[ControllableGeneration]

I assume sometimes it can also be a batchsize problem. Try decrease your batchsize and run again. At least this is true to me when I use cuda 9.0 torch 0.4.1 python 3.6 to run another program.