Closed curiousboy20 closed 3 years ago
Check "torch.cuda.is_available()". If it gives your false, you have to reinstall PyTorch with the correct Cuda.
Check "torch.cuda.is_available()". If it gives your false, you have to reinstall PyTorch with the correct Cuda.
Hi, thank you for the response. I tried to print "torch.cuda.is_available()" and the result is "True" and the problem still exist
Check "torch.cuda.is_available()". If it gives your false, you have to reinstall PyTorch with the correct Cuda.
Hi, thank you for the response. I tried to print "torch.cuda.is_available()" and the result is "True" and the problem still exist
Might be due to the cuda version. I am running spconv 1.0 with following settings; pytorch == 1.0.0 cudaversion == 10.0
@curiousboy20 I got the same error. How did you solve the issue?
@curiousboy20 Anyone solved this issue?
I got the same error.
I too got the same error. has anyone solved it?
I too got the same error. and i just change it into : import spconv.utils and then ingore it ,it can run but later training will have problems
so anyone have solved this issue?
Hi @zhangzirui1998,
Try renaming non_max_suppression with non_max_suppression_cpu. It will work.
Hi @zhangzirui1998,
Try renaming non_max_suppression with non_max_suppression_cpu. It will work.
The file is second/core/non_max_suppression/nms_gpu.py. https://github.com/traveller59/spconv/issues/431#issuecomment-1027717472
Hi, I have installed spconv and followed the instruction with details: Python 3.8.11 Pytorch 1.4 spconv 1.2.1 CUDA 11.4 Ubuntu 18.04 Numba 0.53.1
but when I run the script
python create_data.py kitti_data_prep --data_path=data/sets/kitti_dataset
it said:
I run:
python -c 'import spconv'
and there is no error. Reinstalled the spconv also and it said:
Any help appreciated. Thank you
-------------- update ---------------
I have tried with spconv 1.0 but the same error still exists. Please help