Installing Pytorch on Jetson Orin NX problem

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin NX
1

I tried to install Pytorch 2.8 on my Jetson Orin NX but met some problems.

After some google I download a wheel from here: https://pypi.jetson-ai-lab.io/jp6/cu126/+f/62a/1beee9f2f1470/torch-2.8.0-cp310-cp310-linux_aarch64.whl#sha256=62a1beee9f2f147076a974d2942c90060c12771c94740830327cae705b2595fc

But after I installed it and tried to use it (and even just a import), what I got is an error:

python -c "import torch;print(torch.__version__)"

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/nvidia/zhf/test/venv/lib/python3.10/site-packages/torch/__init__.py", line 2094, in <module>
    _C._initExtension(_manager_path())
  File "/home/nvidia/zhf/test/venv/lib/python3.10/site-packages/torch/cuda/__init__.py", line 364, in <module>
    AcceleratorError = torch._C.AcceleratorError
AttributeError: module 'torch._C' has no attribute 'AcceleratorError'

I don’t know how to solve it, please help, any suggestion is welcome.

5

Hi,

What is your JP version?

Thanks

6

Jetpack version is 6.2.

7

Hi,

Please try using the legacy prebuilt wheel available at:

GitHub - davidl-nv/torch

Thanks.

8

I tried this wheel, unfortunately got the same result:

python -c "import torch"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/nvidia/zhf/test/venv/lib/python3.10/site-packages/torch/__init__.py", line 2079, in <module>
    _C._initExtension(_manager_path())
  File "/home/nvidia/zhf/test/venv/lib/python3.10/site-packages/torch/cuda/__init__.py", line 364, in <module>
    AcceleratorError = torch._C.AcceleratorError
AttributeError: module 'torch._C' has no attribute 'AcceleratorError'
9

Hi,

Please use virtualenv as your virtual environment instead.
You can install and activate it with the following commands:

$ sudo apt-get install virtualenv
$ python3 -m virtualenv -p python3 <chosen_venv_name>
$ source <chosen_venv_name>/bin/activate

Thank you.

10

Thank you. I tries the virtualenv way, and result is the same:

nvidia@nvidia-desktop:~/zhf/test$ python3 -m virtualenv -p python3 v
created virtual environment CPython3.10.12.final.0-64 in 596ms
  creator CPython3Posix(dest=/home/nvidia/zhf/test/v, clear=False, no_vcs_ignore=False, global=False)
  seeder FromAppData(download=False, pip=bundle, setuptools=bundle, via=copy, app_data_dir=/home/nvidia/.local/share/virtualenv)
    added seed packages: pip==25.3, setuptools==80.9.0
  activators BashActivator,CShellActivator,FishActivator,NushellActivator,PowerShellActivator,PythonActivator

nvidia@nvidia-desktop:~/zhf/test$ source v/bin/activate

(v) nvidia@nvidia-desktop:~/zhf/test$ pip install --no-cache torch-2.8.0-cp310-cp310-linux_aarch64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple

...
Installing collected packages: mpmath, typing-extensions, sympy, networkx, MarkupSafe, fsspec, filelock, jinja2, torch
Successfully installed MarkupSafe-3.0.3 filelock-3.20.0 fsspec-2025.10.0 jinja2-3.1.6 mpmath-1.3.0 networkx-3.4.2 sympy-1.14.0 torch-2.8.0 typing-extensions-4.15.0

(v) nvidia@nvidia-desktop:~/zhf/test$ python -c "import torch"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/nvidia/zhf/test/v/lib/python3.10/site-packages/torch/__init__.py", line 2079, in <module>
    _C._initExtension(_manager_path())
  File "/home/nvidia/zhf/test/v/lib/python3.10/site-packages/torch/cuda/__init__.py", line 364, in <module>
    AcceleratorError = torch._C.AcceleratorError
AttributeError: module 'torch._C' has no attribute 'AcceleratorError'
11

Thank you for your time. Finally I think I found the reason. My virtual env use the incorrect pytorch.so (the one in the main env and version is 2.5.0). After I modify the LD_LIBRARY_PATH it is ok.
Thank you again, David.