Easy Setup with Docker and DevContainer for MLLM

September 4, 2025 · View on GitHub

To simplify developers' experience with MLLM, we provide ready-to-use Dockerfile and DevContainer configurations.

1. Using Dockerfile

git clone https://github.com/UbiquitousLearning/mllm.git
cd mllm/docker

# CPU
docker build -t mllm_arm -f Dockerfile.arm .
docker run -it --cap-add=SYS_ADMIN --network=host --cap-add=SYS_PTRACE --shm-size=4G --security-opt seccomp=unconfined --security-opt apparmor=unconfined --name mllm_arm_dev mllm_arm bash

# NVIDIA GPU. Chose your CUDA version: Dockerfile.cuxxx
docker build -t mllm_cu124 -f Dockerfile.cu124 .
docker run -it --gpus all --cap-add=SYS_ADMIN --network=host --cap-add=SYS_PTRACE --shm-size=4G --security-opt seccomp=unconfined --security-opt apparmor=unconfined --name mllm_cu124_dev mllm_cu124 bash

Important Notes:

  1. Dockerfile.arm includes NDK downloads. By using this image, you agree to NDK's additional terms.
  2. QNN SDK contains proprietary licensing terms. We don't bundle it in Dockerfile.qnn - please configure QNN SDK manually.

2. Using DevContainer

To set up with VS Code Dev Containers:

  1. Install prerequisites:

    • Docker
    • VS Code
    • Dev Containers extension

  2. Clone repository with submodules:

    git clone --recursive https://github.com/UbiquitousLearning/mllm.git

  3. Open project in VS Code:

    code mllm

  4. When prompted:

    "Folder contains a Dev Container configuration file. Reopen in container?" Click Reopen in Container

    (Alternatively: Press F1 → "Dev Containers: Reopen in Container")

The container will automatically build and launch with:

  • All dependencies pre-installed
  • Correct environment configuration
  • Shared memory and security settings applied