Based on ubuntu:22.04 (CPU) or nvidia/cuda:12.6.3-base-ubuntu22.04 (GPU).
Other base OS variants: Dockerfiles for Arch Linux and openSUSE Leap are also available if you need a different base distribution.
Pre-built images: CPU and GPU images are automatically built and published to GHCR on every push to
mastervia the CI workflow. Each image is a multi-platform manifest coveringlinux/amd64andlinux/arm64. Pull them directly with:docker pull ghcr.io/swiss-ai/mmore:edge-gpu docker pull ghcr.io/swiss-ai/mmore:edge-cpu
Note: The default target architecture matches the build host. Pass
--platform=<value>to override:
linux/amd64— x86_64 servers (e.g. RCP)linux/arm64— ARM64 machines (e.g. Apple Silicon)
GPU (default):
sudo docker build -f docker/ubuntu/Dockerfile -t mmore .CPU-only:
sudo docker build -f docker/ubuntu/Dockerfile --build-arg DEVICE=cpu -t mmore:cpu .Custom extras (overrides the default --extra all --extra cu126 or --extra all --extra cpu):
sudo docker build -f docker/ubuntu/Dockerfile --build-arg UV_OVERRIDE="--extra process --extra rag --extra cpu" -t mmore .Custom user UID/GID (e.g. for RCP):
sudo docker build -f docker/ubuntu/Dockerfile --build-arg USER_UID=$(id -u) --build-arg USER_GID=$(id -g) -t mmore .# GPU
sudo docker run --gpus all -it -v ./examples:/app/examples -v ./.cache:/home/mmoreuser/.cache mmore
# CPU-only
sudo docker run -it -v ./examples:/app/examples -v ./.cache:/home/mmoreuser/.cache mmore:cpu