I am pretty new to using docker and docker with nvidia. I have a docker image built called 'pixel2mesh_service' which builds from the latest pytorch image running cuda-11.7. I am attempting to run the docker container using my built image docker run --rm --name pixel2mesh_service -it --gpus=all pixel2mesh_service, though without success.
I run into this error: nvidia-container-cli: initialization error: load library failed: libnvidia-ml.so.1: cannot open shared object file: no such file or directory: unknown
Others have posted about this issue though there doesn't seem to be a unique solution. Can the libnvidia-ml.so.1 library be manually installed in the Dockerfile? Are there required nvidia or cuda dependencies that need to be included in the Dockerfile in order to run the container?
Dockerfile:
FROM pytorch/pytorch:latest
WORKDIR /usr/app/
RUN apt-get update && apt-get install -y git libgl1 libegl1 libgomp1 python3-pip wget
COPY requirements.txt .
RUN pip install -r requirements.txt