NuRec GA container nvcr.io/nvidia/nre/nre-ga:26.02.00 fails at startup with missing internal module

Omniverse Core Platform NuRec
1

Hi NVIDIA team,

I am testing the NuRec GA container on Ubuntu 22.04 LTS and the official GA image fails before command parsing with a missing internal module.

Environment:

  • Host OS: Ubuntu 22.04 LTS
  • GPU: 2x RTX 4090
  • Driver: 550.107.02
  • CUDA reported by nvidia-smi: 12.4
  • Current date of test: April 14, 2026

Image:

Repro 1:
docker run --rm -it nvcr.io/nvidia/nre/nre-ga:26.02.00 bash

Observed:
No CUDA runtime is found, using CUDA_HOME=‘/usr/local/cuda’
ModuleNotFoundError
To identify missing module enable ‘debug: True’ in the relevant config in scripts/pycena/configs
For additional information consult scripts/pycena/README.md
No module named ‘11llI111IlI1ll1IlI11l11I1’

Repro 2:
docker run --rm -it --gpus all nvcr.io/nvidia/nre/nre-ga:26.02.00 bash

Observed:
ModuleNotFoundError
To identify missing module enable ‘debug: True’ in the relevant config in scripts/pycena/configs
For additional information consult scripts/pycena/README.md
No module named ‘11llI111IlI1ll1IlI11l11I1’

This shows the failure is not caused by missing GPU access, because adding --gpus all removes the CUDA warning but the same module import error remains.

Repro 3:
docker run --rm -it --gpus all --entrypoint /bin/bash nvcr.io/nvidia/nre/nre-ga:26.02.00

Inside the container I checked:
find /app -type f \( -name ‘11llI111IlI1ll1IlI11l11I1’ -o -name ‘lI1II1I1II1lIl11l1l11lII1’ \)
ls -la /app/internal/scripts/pycena
find /app/internal/scripts/pycena/runtime/pycena_run.runfiles/_main/internal/scripts/pycena -maxdepth 3 \( -name README.md -o -type d -name configs \)

Observed:

  • The missing internal module files are not present in the image.
  • /app/internal/scripts/pycena only contains runtime/
  • The paths referenced by the error message (scripts/pycena/README.md and scripts/pycena/configs) are not present in the image either.

Additional notes:

  • I removed the local image and pulled it again.
  • The pulled digest matches the official NGC digest exactly.
  • I can pull nre-ga successfully, so this does not appear to be a local Docker login issue.

Could you please confirm whether this GA image is currently broken or if there is a corrected tag/digest that should be used instead?

Thank you.

2

Hi,
The GA is not broken. The container in this initial version is obfuscated, that is why we can’t get enough details on the nature of the error. When you see ModuleNotFoundError is because the code hit an unexpected path. So there must be an error in the arguments or input data provided. I would suggest you start from scratch (clearing any files previously generated) and follow the documentation examples carefully. Can you provide us more detail about the workflow you are running?
has your data being converted correctly into NCore?
are you referencing the right folders and files when running the containers?

Respectfully,
Mauricio.