Tlt-streamanalytics training in Singularity


Hi ! Is there someone that achieved to properly run an application sample like Faster RCNN in the docker with Singularity without root access ?
When I pull then run the .sif file with these commands :
singularity pull tlt-streamanalytics:v3.0-dp-py3.sif docker://
singularity run tlt-streamanalytics:v3.0-dp-py3.sif
This pops :

INFO: Could not find any nv files on this host!
mkdir: cannot create directory ‘/opt/ngccli’: Read-only file system

Then I have no examples or tlt-experiments folders.

After many local installations of ngccli, pip and tlt in my ~/.local/bin folder, I was able to reach the step before the conversion of the kitty dataset.

Any idea ?
Thanks !

Please follow TLT Launcher — Transfer Learning Toolkit 3.0 documentation and
Getting Started With TLT — Transfer Learning Toolkit 3.0 documentation

Dear Morgan,

I’m sorry but I can’t find any information about running TLT with Singularity without root privileges. Is it possible ?
Only these softwares are installed on the cluster :

Thanks for your answer.

Official TLT is not verified with Singularity. We do not know its status.
More, why will you run TLT with Singularity?

Because it’s the only docker alternative I have on this cluster, and as I understood a docker is needed to run TLT. But if there’s a way to install TLT and make it run without root privileges, I’m really interested. (I can install softwares from source in my local HOME directory)

See below “post-installation steps”, it will help you.

If you have followed the default installation instructions for docker-ce you may need to have sudo access to run docker commands. In order to circumvent this, TLT recommends you to follow these post-installation steps to make sure that the docker commands can be run without sudo.

From Requirements and Installation — Transfer Learning Toolkit 3.0 documentation