Error running setup on torch2trt

Autonomous Machines Jetson & Embedded Systems Jetson Nano
#1

Sorry for this simpleton problem but I am ramping up on Linux, Python, and all of the above to develop an application for work ( while also doing my day job) …

I am trying to install the torch2trt python module.
I have Pytorch already on my Jetson Nano ( with the AI build)
I follow this procedure:

git clone GitHub - NVIDIA-AI-IOT/torch2trt: An easy to use PyTorch to TensorRT converter
cd torch2trt
python setup.py install

But I get the following error:

Traceback (most recent call last):

  • File “setup.py”, line 2, in *
  • import torch*
    ImportError: No module named torch

I am new to this environment so I’m sure it’s a moment of stupidity on my part. Please enlighten me.

BTW: I searched for torch module and it shows up in my usr/local/lib/python3.6/dist-packages directory so it’s there. The install went to my home directory. Does this have to do with my intall being in a directory far far away from where Python wants it?

#3

Hi,

Sorry that the document doesn’t update correctly.
You will need to use python3.6 for the installation:

python3 setup.py install

Thanks.

#4

I figured it out a little later, got through that hurdle and now I am looking at :
OSError: libcurand.so.10:

When I try to verify the install with >>>import torch

Now what am I missing?

#5

Hi,

Please check this topic for more information about the error:

Thanks.