Closed XinYu-Andy closed 1 year ago
Can you try installing with pytorch 1.13? Use the commands on the tab that says Torch 1.13.1 with CUDA 11.7 here: https://docs.nerf.studio/en/latest/quickstart/installation.html. Make sure to create a conda environment first using the instructions at the top of the page.
Ah also, the instructions are a bit confusing. From the installation page, you should only follow the commands up to tinycudann. Then, come back to our readme and go to step 2. The instructions have been updated to be more clear.
May I ask what version of PyTorch you are using? I found that my torch version became 2.0 after following these steps to configure the environment, and as a result, I am unable to successfully import tinycudann.
pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 -f https://download.pytorch.org/whl/torch_stable.html pip install ninja git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch pip install nerfstudio
It will report an error if I want to train the second stage (edit part), i.e.:ImportError: libtorch_cuda_cu.so: cannot open shared object file: No such file or directory