GitLab ML CLI

A command-line interface tool for managing machine learning models in GitLab's Model Registry. This tool provides a seamless experience for ML teams to version, track, and deploy their models using GitLab's infrastructure.

Features

• Model Registry Management:
  • Create, delete, and list models
  • Manage model metadata and tags
  • Track model lineage and dependencies
• Model Versioning:
  • Upload new model versions with semantic versioning
  • Download specific versions for deployment
• Access Control:
  • GitLab authentication and authorization

Installation

Using pip

pip install gitlab-ml

Using Poetry

poetry add gitlab-ml

From Source

git clone https://github.com/ahmetoner/gitlab-ml.git
cd gitlab-ml
poetry install

Configuration

Configuration File

Create a configuration file at ~/.config/gitlab-ml/config.yml:

gitlab:
  # GitLab instance URL
  url: "https://gitlab.com"
  
  # Personal access token with api scope
  token: "your-personal-access-token"
  
  # Default project for model registry
  default_project: "group/project"

Environment Variables

Alternative to config file, you can use environment variables:

# Required settings
export GITLAB_ML_TOKEN="your-personal-access-token"
export GITLAB_ML_URL="https://gitlab.com"
export GITLAB_ML_PROJECT="group/project"

Usage

Model Management

# List all models with different output formats
gitlab-ml models list
gitlab-ml models list --format json
gitlab-ml models list --format yaml

# Create a new model
gitlab-ml models create my-model \
    --description "My ML model"

# Delete a model (with confirmation)
gitlab-ml models delete my-model

# Force delete without confirmation
gitlab-ml models delete my-model --force

Version Management

# Upload a new model version
gitlab-ml models upload my-model 1.0.0 ./model.pkl

# Upload a directory of model artifacts
gitlab-ml models upload my-model 1.0.0 ./model_dir

# Download a specific version
gitlab-ml models download my-model 1.0.0 \
    --output ./models/

Development

1. Clone the repository:

   git clone https://github.com/ahmetoner/gitlab-ml.git
   cd gitlab-ml

2. Install development dependencies:

   poetry install --with dev

3. Activate virtual environment:

   poetry shell

4. Run tests:

   # Run all tests
   poetry run pytest
   
   # Run with coverage
   poetry run pytest --cov=gitlab_ml
   
   # Run specific test file
   poetry run pytest tests/test_models.py

5. Code formatting and linting:

   # Format code
   poetry run black .
   poetry run isort .
   
   # Run linters
   poetry run mypy .
   poetry run ruff .

Troubleshooting

Common Issues

1. Authentication Errors

   Error: GitLab authentication failed
   

   • Verify your token has the correct permissions (api scope)
   • Check if token is expired
   • Ensure GitLab URL is correct

2. Project Access

   Error: Project not found or no access
   

   • Verify project path is correct
   • Check if you have sufficient permissions
   • Ensure project exists and is accessible

3. Upload Failures

   Error: Failed to upload model version
   

   • Check file permissions
   • Verify disk space
   • Ensure version follows semver format

API Documentation

Python API

from pathlib import Path
from gitlab_ml.api.client import get_gitlab_client, GitLabClient
from gitlab_ml.api.models import ModelRegistry

# Initialize client with custom URL and token
client = GitLabClient(
    url="https://gitlab.com",
    token="glpat-xxxxxxxxxxxx",
    default_project="group/project"
)
registry = ModelRegistry(client)

# Or use environment variables
import os
os.environ["GITLAB_ML_URL"] = "https://gitlab.com"
os.environ["GITLAB_ML_TOKEN"] = "glpat-xxxxxxxxxxxx"
os.environ["GITLAB_ML_PROJECT"] = "group/project"
client = get_gitlab_client()
registry = ModelRegistry(client)

# List models
models = registry.list_models()

# Upload model
registry.upload_version(
    model_name="my-model",
    version="1.0.0",
    path=Path("./model.pkl")
)

# Download model
registry.download_version(
    model_name="ml-model-new",
    version="1.2.2",
    output_dir=Path("./downloaded_model")
)

REST API Integration

The CLI integrates with GitLab's REST and GraphQL APIs. For custom integrations, refer to:

• GitLab REST API Documentation
• GitLab GraphQL API Documentation

Contributing

We welcome contributions! Please follow these steps:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Run tests and linting
5. Submit a merge request

License

This project is licensed under the MIT License - see the LICENSE file for details.