Has anyone tried not to use a container environment?

[PamelaDDD]

I encountered the following error on GPU 3090, but it can run successfully on Titan Xp. I suspect that the Cuda version in the container is relatively low. So has anyone tried to build an environment to run uniter without using the official container environment? Or does anyone have a better solution?

[2292384454]

Hello, I have encountered a problem with you on 3090, and Hope my method can help you. ①You should install a newer CUDA in your container than CUDA10.0. I installed CUDA 11.3 from https://developer.nvidia.com/. ②And then you need re-installe your torch in container . I did this like "pip install torch==1.10.0+cu113 torchvision==0.11.1+cu113 torchaudio==0.10.0+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html". ③Then you will encounter the problem of "ImportError: cannot import name 'container_abcs'". You can solve that by reading"https://blog.csdn.net/qq_19313495/article/details/120361059".