train.py : ImportError: libtorch_cuda_cu.so: cannot open shared object file: No such file or directory

[bring-nirachornkul]

Torch version: 1.9.1.post3
CUDA version: 11.1
Python Version : 3.7
Pytorch3D : 0.7.6

Hardware :
CPU : AMD Ryzen 9 5900X 12-Core Processor
GPU : RTX 3090
RAM : 64 GBs
Linux OS : Ubuntu 23.04

When I run python train.py then this error occurred :

(v2a) blinkdrive@blinkdrive-System-Product-Name:~/Documents/Projects/Summer2024/02vid2avatar/code$ python train.py
Traceback (most recent call last):
  File "train.py", line 1, in <module>
    from v2a_model import V2AModel
  File "/home/blinkdrive/Documents/Projects/Summer2024/02vid2avatar/code/v2a_model.py", line 3, in <module>
    from lib.model.v2a import V2A
  File "/home/blinkdrive/Documents/Projects/Summer2024/02vid2avatar/code/lib/model/v2a.py", line 4, in <module>
    from .deformer import SMPLDeformer
  File "/home/blinkdrive/Documents/Projects/Summer2024/02vid2avatar/code/lib/model/deformer.py", line 4, in <module>
    from pytorch3d import ops
  File "/home/blinkdrive/miniconda3/envs/v2a/lib/python3.7/site-packages/pytorch3d/ops/__init__.py", line 7, in <module>
    from .ball_query import ball_query
  File "/home/blinkdrive/miniconda3/envs/v2a/lib/python3.7/site-packages/pytorch3d/ops/ball_query.py", line 10, in <module>
    from pytorch3d import _C
ImportError: libtorch_cuda_cu.so: cannot open shared object file: No such file or directory

[MoyGcc]

Hi,

It's an installation problem of PyTorch3D. Could you double check that you have installed it successfully (with CUDA enabled)? You can refer to https://github.com/facebookresearch/pytorch3d/blob/main/INSTALL.md for detailed instructions.

[bring-nirachornkul]

Hi,

It's an installation problem of PyTorch3D. Could you double check that you have installed it successfully (with CUDA enabled)? You can refer to https://github.com/facebookresearch/pytorch3d/blob/main/INSTALL.md for detailed instructions.

I have to these steps already. Here is the error after run train.py

(v2a) blinkdrive@blinkdrive-System-Product-Name:~/Documents/Projects/Summer2024/02vid2avatar/code$ python train.py
Traceback (most recent call last):
  File "train.py", line 1, in <module>
    from v2a_model import V2AModel
  File "/home/blinkdrive/Documents/Projects/Summer2024/02vid2avatar/code/v2a_model.py", line 3, in <module>
    from lib.model.v2a import V2A
  File "/home/blinkdrive/Documents/Projects/Summer2024/02vid2avatar/code/lib/model/v2a.py", line 4, in <module>
    from .deformer import SMPLDeformer
  File "/home/blinkdrive/Documents/Projects/Summer2024/02vid2avatar/code/lib/model/deformer.py", line 4, in <module>
    from pytorch3d import ops
  File "/home/blinkdrive/miniconda3/envs/v2a/lib/python3.7/site-packages/pytorch3d/ops/__init__.py", line 5, in <module>
    from .graph_conv import GraphConv
  File "/home/blinkdrive/miniconda3/envs/v2a/lib/python3.7/site-packages/pytorch3d/ops/graph_conv.py", line 8, in <module>
    from pytorch3d import _C
ImportError: libcudart.so.10.1: cannot open shared object file: No such file or directory

[MoyGcc]

Could you refer to the previous comment #49 (comment)? It's basically the unsuccessful installation of PyTorch3D (with CUDA). You can feel free to change the python/torch/cuda versions to match some prebuilt wheels of PyTorch3D which should ensure a bug-free installation.