- clone this repo as a submodule in the `.devcontainer` folder
- configure the `.devcontainer.json` to build the desired docker file
- alternatively, use 'image': 'gitlab.ilabt.imec.be:4567/tlips/gpulab/jupyterlab-pytorch:latest' to avoid the need for submodules if you do not want to be able to alter the container. (no auth?)
To enable GPU on local devices: install the [nvidia docker toolkit](https://github.com/NVIDIA/nvidia-docker) and pass the `--gpus all` arg to docker to allow for discovering the GPUs. The nvidia toolkit will use the on-device drivers while installing the cuda toolkit in the container. To validate if the gpus are enabled, run the `nvidia-smi` command in the container.
## IDLab JupyterHub:
- use `gitlab+deploy-token-8:YHsqg1MHW-C4GJpr_B-D@gitlab.ilabt.imec.be:4567/tlips/gpulab/jupyterlab-pytorch:latest` as custom docker image.