Visualizing "ML class" failes after supervised classifier

I am doing a classification of a Sentinel 2 image (using the tutorial from RUS:
https://rus-training.eu/training/lithological-classification-with-sentinel-1-sentinel-2, and https://www.youtube.com/watch?v=iJvW9q4v78w&t=100s).
After creating new vector data container for the supervised max. likelihood classification, SNAP failes to show the file correctly when reopening it (see screenshot). This only occured after opening it a few times without any problems.
Also, some files are not presented in the image viewer when others are opened.

I am working on a macbook pro and did not move or delete any files.
Appreciate any help!!

Can you please show that image before classification (including the vectors)?

This is the resampled and reprojected subset of a S2 image. With this, I did the supervised classification by drawing the polygons.

Just curious, why did you reduce the Sentinel-2 stack to the visual bands? You could get way better results with the full multi-specral information.

I am still wondering where the screenshot image is from: You cannot display an RGB image of the original bands after the classification (because only class number and confidence are left).

Sorry for the late answer. I am very new to SNAP and trying to figure out what fits the best for my research. I am doing a classification of a coastline (Lithology, Sand, Rock, Dunes, Vegetation etc.). Its quite hard due to vegetation cover and buildings, but we’ll see.
I started over again and for some reason everything worked without any problems. Probably just some saving/location error.

thank you for reporting - happy to hear that you solved your problem! Good luck with the classification - let us know if you encounter any other problems.