Running the TLT on a Jetson Nano

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

To preface, I have little experience with the Jetson Nano and Nvidia’s software, but I was given a task that requires the use of Nvidia’s TLT. I was following the directions of this guide https://docs.nvidia.com/metropolis/TLT/tlt-getting-started-guide/index.html to setup the TLT on the Jetson Nano and when I went to enter this command:

docker run --runtime=nvidia -it -v /home/tyler/tlt-experiments:/workspace/tlt-experiments nvcr.io/nvidia/tlt-streamanalytics:2.0_py3 /bin/bash

the following error occurs:

standard_init_linux.go:211: exec user process caused “exec format error”

Research into the error leads me to the following thread of this forum: Docker run error - "exec format error"

Reading through the thread it seems like the TLT can’t be run on the Jetson Nano itself and relies on a separate host machine to train models on? Is there a way to make this work solely on the Jetson Nano? Any help regarding this problem is appreciated.

2

The docker should work on host PC.
After training, the etlt model or trt engine can be deployed into Jetson Nano.

3

The requirements in the TLT documentation Integrating TAO Models into DeepStream — TAO Toolkit 3.22.05 documentation suggest that the training be done on the Ubuntu OS. For Windows users, would using a virtual machine running Ubuntu be able to run the training portion of the TLT?

4

@LearningGuy
It is expected to work.