(py)torch error: Torch not compiled with CUDA enabled

DRIVE OS Version: 7.0.3

Issue Description:

Python torch support installed via

pip install torch

Tried to initialize torch

import torch
_ = torch.randn(1, device=‘cuda’)

But there is an error

File “/home/compal/miniconda3/envs/cdat/lib/python3.13/site-packages/torch/cuda/init.py”, line 417, in _lazy_init
raise AssertionError(“Torch not compiled with CUDA enabled”)

Dear @dan_borlovan
PyTorch is not officially supported on DRIVE platform. We suggest you to use ONNX - > TensorRT path for optimal performance of models.

We use (py)torch to execute some matrix calculations on gpu

(For inference we use pycuda)