RuntimeError: cuda runtime error (38) : no CUDA-capable device is detected at /opt/conda/conda-bld/pytorch_1532584813488/work/aten/src/THC/THCGeneral.cpp:74

[guohaoyuan]

Thank you for this code, it's amazing! I've been learning about your project recently, but this error occurred at the beginning of the training. I thought the graphics card didn't match the driver. After trying it out, I found that it wasn't. ################################################## Loaded dataset voc_2007_trainval for training Set proposal method: gt Appending horizontally-flipped training examples... voc_2007_trainval gt roidb loaded from /home/ghy/faster-rcnn.pytorch-master/data/cache/voc_2007_trainval_gt_roidb.pkl done Preparing training data... done

---

before filtering, there are 10022 images... after filtering, there are 10022 images... 10022 roidb entries THCudaCheck FAIL file=/opt/conda/conda-bld/pytorch_1532584813488/work/aten/src/THC/THCGeneral.cpp line=74 error=38 : no CUDA-capable device is detected Traceback (most recent call last): File "trainval_net.py", line 222, in im_data = im_data.cuda() RuntimeError: cuda runtime error (38) : no CUDA-capable device is detected at /opt/conda/conda-bld/pytorch_1532584813488/work/aten/src/THC/THCGeneral.cpp:74

############################################### Does anyone know how to solve it?

[guohaoyuan]

It has been solved. Reasons for not calling CUDA!

[yaelbrumer]

Hi, I'm still seeing the same error. What was the solution for that?

[guohaoyuan]

Hi, I'm still seeing the same error. What was the solution for that? 我的原因在：命令行需要修改GPU的参数，即，调用哪个GPU

[kangkang59812]

@guohaoyuan --cuda cuda:0 ?

[gmt710]

One possible solution is to put the following two statements at the front of the code(trainval_net.py).

import os
os.environ["CUDA_VISIBLE_DEVICES"] = '0'

[zzu0654]

I'm still seeing the same error.

[ShaheenPerveen]

Hi I am also encountering similar issue, even after trying all the above solves. @zzu0654 were you able to solve it?

[ahtsham58]

Also getting the same error. My GPU is cross-verified and its active.

you can see the last thread here. https://github.com/rusty1s/pytorch_geometric/issues/2489