A collection of devcontainer configurations I usually use for my deep learning / machine learning projects.
Important
For GPU passthrough to work in some configurations, you'll need to install
the nvidia-container-toolkit first. Please refer to the official docs.
Install the devcontainer-cli via npm
npm install -g @devcontainers/cli
# you may have to use sudo in some linux distros
# especially if npm installed as a system packageThen run the configuration you want by
devcontainer up --workspace-folder <path-to-configuration>
# example
devcontainer up --workspace-folder ./torch-uvIf you only wish to build a devcontainer to run it later,
devcontainer build --workspace-folder <path-to-configuration>
# example
devcontainer build --workspace-folder ./jax-uvImportant
These configurations are for Nvidia GPUs and x86_64 based systems. Besides, GPU passthrough isn't there for macOS/Asahi Linux running Apple Silicon yet.
Once up and running, you can connect to the devcontainer through VS Code or Jetbrains IDEs (these are the ones I use actually!).