Nano LLM container error

We are running jet pack 6.0. Trying build a custom container with the following docker file,
Docker file


FROM dustynv/nano_llm:humble-r36.3.0

ENV DEBIAN_FRONTEND=noninteractive

Set the LD_LIBRARY_PATH environment variable

ENV LD_LIBRARY_PATH=/usr/lib/aarch64-linux-gnu/nvidia:$LD_LIBRARY_PATH

Install system dependencies

RUN apt-get update && apt-get install -y --no-install-recommends
libportaudio2
libasound2-dev
pulseaudio
alsa-utils
libsdl2-dev
libsdl2-mixer-2.0-0
portaudio19-dev
git
&& rm -rf /var/lib/apt/lists/*

Install additional Python dependencies

RUN pip3 install --no-cache-dir --ignore-installed
python-dotenv
websockets
pygame
termcolor
pyaudio

Set the working directory for your application

WORKDIR /app

Create logs directory

RUN mkdir -p /app/logs && chmod 777 /app/logs

Copy your Python scripts into the container

COPY nano-llm-detection.py audio_manager.py websocket_server.py .env ./

Set the entrypoint to run your script

ENTRYPOINT [“python3”, “nano-llm-detection.py”]


The build is successful. But When I try to run the image with the command,

docker run --runtime nvidia
–network host
–device /dev/video0:/dev/video0
–device /dev/snd:/dev/snd
-v /home/checkout/Downloads/detections/.env:/app/.env
-v /home/checkout/media/background_music:/app/media/background
-v /home/checkout/media/greetings:/app/media/store
-v /tmp/.X11-unix:/tmp/.X11-unix
-v /run/user/$(id -u)/pulse:/run/pulse
-e DISPLAY=$DISPLAY
-e PULSE_SERVER=unix:/run/pulse/native
–group-add $(getent group audio | cut -d: -f3)
–user $(id -u):$(id -g)
custom-nano-llm-detection:latest

Im getting this following module import error.

Traceback (most recent call last):
File “/app/nano-llm-detection.py”, line 18, in
from nano_llm import NanoLLM, ChatHistory, remove_special_tokens
File “/opt/NanoLLM/nano_llm/init.py”, line 2, in
from .nano_llm import NanoLLM
File “/opt/NanoLLM/nano_llm/nano_llm.py”, line 15, in
from .vision import CLIPVisionModel, TIMMVisionModel, MMProjector
File “/opt/NanoLLM/nano_llm/vision/init.py”, line 3, in
from .clip import CLIPVisionModel, TIMMVisionModel
File “/opt/NanoLLM/nano_llm/vision/clip.py”, line 3, in
from clip_trt import CLIPVisionModel, TIMMVisionModel
ModuleNotFoundError: No module named ‘clip_trt’

This section of the import statements was copied directly from vision/video.py example. I would think that dustynv/nano_llm:humble-r36.3.0 would already have the needed modules like clip_trt. Why am i still geting this import failure error?

There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

Hi,

To narrow down the issue, could you try to use the dustynv/nano_llm:humble-r36.3.0 to see if the same error occurs?
Please launch the container and try the vision/video.py.

Thanks.