Downloadable environment¶
About¶
This image enables new users to get started with ML-Git in a lightweight Linux-based image without worrying about configurations. The image also include a git repository with a predefined dataset and a minio instance populated with the dataset's data.
How to use:¶
-
Ensure that you have Docker installed.
-
Inside root of ML-Git directory build the image locally with the following command:
make docker.build
or
docker build -t mlgit_docker_env -f docker/Dockerfile .
- Run the Docker container to launch the built image:
make docker.run
or
docker run -it -p 8888:8888 --name mlgit_env mlgit_docker_env
Port 8888 will be used to start the jupyter notebook web service.
Using the ML-Git with environment (inside docker container):¶
The container has an ML-Git project initialized inside the directory workspace, the content of the versioned tag is an image from mnist database.
You can execute the command checkout directly to tag:
ml-git datasets checkout handwritten__digits__mnist__1
Summary of files in image:¶
local_server.git (local git repository, used to store metadafiles).
data (directory used by the bucket to store project data).
init.sh (script that run basic command to use ml-git).
minio (minio executable).
local_ml_git_config_server.git (local git repositoy with configuration files, used by ml-git clone).
ml-git (source code of ml-git).
workspace (initialized ml-git project).