Skip to main content

GitLab MLOps Python Client

A Python client for integrating GitLab's MLOps features, specifically designed to leverage GitLab's experiment tracking and model registry capabilities. This client simplifies the management of machine learning workflows by combining MLflow's flexibility with GitLab's powerful MLOps tools.

Build Status Latest Release

Features

  • GitLab Experiment Tracking: Easily track machine learning experiments within GitLab.
  • Model Registry Integration: Register and manage models in GitLab's model registry.
  • Experiment Management: Create and manage experiments directly from the client.
  • Run Tracking: Initiate and monitor training runs with ease.
  • Model Lifecycle Management: Promote runs to model versions effortlessly.

Feedback

The MLOPs Python client is currently in Beta. We welcome feedback and feature requests either in this repository's issue tracker or on our feedback issue

Installation

Install the client using pip:

pip install gitlab-mlops

Alternatively, to install the latest version from the source, use Poetry:

# Clone the repository
git clone https://gitlab.com/gitlab-org/modelops/mlops/gitlab-mlops.git
cd gitlab-mlops

# Install dependencies
poetry install

Usage

Configuration

Configure the client using environment variables or initialization arguments.

Using Environment Variables:

export MLFLOW_TRACKING_URI="https://gitlab.com/api/v4/projects/<project_id>/mlflow"
export MLFLOW_TRACKING_TOKEN="your_access_token"
from gitlab_mlops import Client

client = Client()

Using Initialization Arguments:

from gitlab_mlops import Client

client = Client(
    tracking_uri="https://gitlab.com/api/v4/projects/<project_id>/mlflow",
    gitlab_token="your_access_token"
)

Create a New Experiment

from gitlab_mlops import Client

client = Client()
experiment = client.create_experiment("Experiment Name")

Create a New Run

from gitlab_mlops import Client

client = Client()
experiment = client.get_experiment(name="Experiment Name")
run = experiment.create_run()

Log Parameters, Metrics, and Model artifacts

from gitlab_mlops import Client

client = Client()

run = client.get_run(run_id="19e547d3-8de7-42c3-9575-fae1142dec69")

run.log_param(key="param_name", value="value")
run.log_metric(key="metric_name", value=0.95)
run.log_artifact(
    local_path="<path to the model file or directory>",
    artifact_path="<optional relative path to log the artifact(s) at>"
)

Create a New Model

from gitlab_mlops import Client

client = Client()
model = client.create_model(name="Model Name", description="Model Description")

Create a New Model Version

from gitlab_mlops import Client

client = Client()

model = client.get_model(name="Model Name")

model_version = model.create_version(description="New version", version="1.0.0")

model_version.log_param(key="param_name", value="value")
model_version.log_text("How to run this model [...]", "README.md")

Importing existing models

To import existing models from various external sources, you can use the export_to_gitlab method. This method allows you to import models from the following sources:

  • Google Cloud Platform - Vertex AI

Importing from Google Cloud Platform - Vertex AI

from gitlab_mlops import Client
from gitlab_mlops.providers.gcp.vertex import VertexAIClient

client = Client()
vertex = VertexAIClient(
    project_id="your_project_id",
    location="us-central1",
    client=client
)

resource_name = "projects/your_project_id/locations/us-central1/models/your_model_id"

vertex.export_model_to_gitlab(
    model_id=resource_name
)

Metadata

Release files for gitlab-mlops 0.3.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for gitlab-mlops 0.3.0
File Size Uploaded Provenance Actions
gitlab_mlops-0.3.0.tar.gz 12.6 kB Without provenance Details for gitlab_mlops-0.3.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for gitlab-mlops 0.3.0
File Interpreter ABI Platform Actions
gitlab_mlops-0.3.0-py3-none-any.whl Python 3 none any Details for gitlab_mlops-0.3.0-py3-none-any.whl

Total release size: 29.9 kB

Release files / gitlab_mlops-0.3.0.tar.gz

Download URL gitlab_mlops-0.3.0.tar.gz
Size 12.6 kB
Tags Source
SHA-256 checksum
How to use checksums
7c5497a8fddcf8dd06f854cfb7db2a84c372cb14c1ba9e54b34cbe41b23bca77
BLAKE2b-256 checksum
How to use checksums
57473bc089963652c73fe96ce9d10414ebb0ed29d70e4f1c82ad2cb3f71cea0a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via poetry/2.1.1 CPython/3.11.9 Linux/5.15.154+

Release files / gitlab_mlops-0.3.0-py3-none-any.whl

Download URL gitlab_mlops-0.3.0-py3-none-any.whl
Size 17.4 kB
Tags Python 3
SHA-256 checksum
How to use checksums
601cce15ed28d4cf4b2c4ded524ce67b31ba17fcd97e951624e2c7110ca148ea
BLAKE2b-256 checksum
How to use checksums
0f7ac21cea0478f7ea0b641cfdf5feb8919374c1c6368d2ca196bd18087c7f46
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via poetry/2.1.1 CPython/3.11.9 Linux/5.15.154+

Security for gitlab-mlops 0.3.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page