I’ve been diving into the “Getting Started with AI on Jetson Nano” course and got my environment all sorted out using Docker. However, I’ve hit a snag – it seems like every time I restart the kernel in JupyterLab or power down my Nano, I end up having to retrain the model for my projects from scratch.

I’m a bit stumped as to why this keeps happening. It feels like either the trained models aren’t persisting properly or there’s an issue with how my environment is set up. I’m not sure where to even begin troubleshooting this.

Here’s what I’ve done so far:

Went through the “Getting Started with AI on Jetson Nano” course of NVIDIA® Deep Learning Institute (DLI)
Got my environment up and running using Docker.

Noticed the annoying need to retrain the model after every kernel restart or device shutdown.
If anyone has any insights or tips on how to tackle this issue, I’d be incredibly grateful. Thanks a bunch!