Install docker and NVIDIA Container Toolkit.

docker build . -t solr-cuvs-demo:25.08

docker run -p 8888:8888 -p 8983:8983 --runtime=nvidia --gpus all solr-cuvs-demo:25.08

In the browser, open the URL printed (starting with http://127.0.0.1:8888/...). Open the solr-cuvs-demo.ipynb notebook and execute the cells one by one. Everytime you update the notebook, copy it back to this directory (demo.ipynb) so that those changes persist in subsequent runs.

1. "CUDA driver version is insufficient for CUDA runtime version": This indicates you didn't start the docker container with the --runtime-nvidia --gpus all

2. "nvidia-container-cli: requirement error: unsatisfied condition: cuda>=12.9, please update your driver to a newer version, or use an earlier cuda container: unknown." during the start of the docker run, might indicate that you have a NVIDIA Driver on your host machine that doesn't support cuda-12.9.