Not able to launch TLT3 training

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

Hi,

  1. I have installed all requirements for TLT3 as per documentation.
  2. Launched virtual environment “launcher” using “workon launcher”
  3. Launched jupyter notebook and opened dssd.ipynb

Followed all the steps and finally when I tried to launch training I get this error
AssertionError: Config path must be a valid unix path. No file found at: /home//.docker/config.json

Is there a way to set the config file path for tlt? or will TLT always expect the config file to be at this path?
If its the latter case, then it will not be possible to launch multiple instances of TLT training on different dockers.

Please clarify

Thanks

3

See Question about running NGC for TLT3.0 - #13 by Morganh

For AssertionError: Config path must be a valid unix path. No file found at: /root/.docker/config.json , please consider below solution.
You need to run docker login nvcr.io

TLT Launcher — Transfer Learning Toolkit 3.0 documentation also mentioned this step.

4

Thanks, yes missed that step.
I tried to login with my NGC email ID and password, that’s not working.

I already have a API key
Does key come in place of username or password?

5

It worked, could login with “$oauthtoken” and my API key as password.
But I still get this error “AssertionError: Config path must be a valid unix path. No file found at: /home//.docker/config.json”
But had to use sudo, hope its fine?
docker works in general without sudo, but “docker login nvcr.io” requires sudo.

6

Please see TLT Launcher — Transfer Learning Toolkit 3.0 documentation

Once you have installed docker-ce, please follow the post-installation steps to make sure that the docker can be run without sudo.

7

Thanks for your support.