PyTorch on L4T Docker image

Autonomous Machines Jetson & Embedded Systems Jetson Nano
1

Hello, I’m geting following error when installing PyTorch using L4T Docker image on Jetson Nano. Here is the Dockerfile. Can you please look into it? Thanks

FROM nvcr.io/nvidia/l4t-base:r32.3.1

ENV DEBIAN_FRONTEND noninteractive
RUN apt-get update && apt-get install -y \
	python3-opencv ca-certificates python3-dev git wget sudo python-setuptools && \
  rm -rf /var/lib/apt/lists/*

ENV PATH="/home/appuser/.local/bin:${PATH}"
RUN wget https://bootstrap.pypa.io/get-pip.py && \
	python3 get-pip.py && \
	rm get-pip.py

RUN wget https://nvidia.box.com/shared/static/phqe92v26cbhqjohwtvxorrwnmrnfx1o.whl -O torch-1.3.0-cp36-cp36m-linux_aarch64.whl
RUN pip3 install numpy torch-1.3.0-cp36-cp36m-linux_aarch64.whl

After that if I do

docker build -t jetson-pytorch:latest .
docker run -it jetson-pytorch:latest bash

root@21592ebf82b0:/# python3
Python 3.6.9 (default, Nov  7 2019, 10:44:02) 
[GCC 8.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.6/dist-packages/torch/__init__.py", line 81, in <module>
    from torch._C import *
ImportError: libnvToolsExt.so.1: cannot open shared object file: No such file or directory
>>> exit()
root@21592ebf82b0:/# exit
exit
2

Hi,

libnvToolsExt.so.1 is located at /usr/local/cuda-10.0/targets/aarch64-linux/lib/.
Could you try to export the path to LD_LIBRARY_PATH and try it again?

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-10.0/targets/aarch64-linux/lib/

Thanks.

3

Thanks for the response. Its not there.

root@69f5ee8f643e:/# printenv | grep LD_
LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64:/usr/local/cuda-10.0/targets/aarch64-linux/lib:
root@69f5ee8f643e:/# ls /usr/local/cuda-10.0/targets/aarch64-linux/lib/
libcudadevrt.a  libcudart_static.a  stubs
4

Actually was able to figure it out but getting this now.

jetson-nano:$ docker run -it --rm --net=host --runtime nvidia  -e DISPLAY=$DISPLAY -v /tmp/.X11-unix/:/tmp/.X11-unix jetson-pytorch:latest bash
root@jetson-nano:/# python3
Python 3.6.9 (default, Nov  7 2019, 10:44:02) 
[GCC 8.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.6/dist-packages/torch/__init__.py", line 81, in <module>
    from torch._C import *
ModuleNotFoundError: No module named 'numpy.core._multiarray_umath'
>>> exit()
5

Hi,

ModuleNotFoundError: No module named 'numpy.core._multiarray_umath'

This is related to the installed numpy version.
Could you check if this command helps?

RUN pip install --upgrade numpy

Thanks.