>MLflow

MLflow

An open-source platform for managing the end-to-end machine learning lifecycle.

jupyter/scipy-notebook

minio/minio

tensorflow/serving

mlflow/mlflow-export-import

What is MLflow?

The mlflow image provides MLflow, an open-source platform for managing the end-to-end machine learning lifecycle. MLflow helps teams track experiments, package ML code into reproducible runs, manage model artifacts, and deploy models into production.This image runs the MLflow CLI or server in a containerized environment, making it easy to integrate with CI/CD pipelines, Kubernetes workloads, or local experimentation. It’s commonly used to centralize experiment tracking (mlflow tracking server), serve models via REST, or store and retrieve artifacts using local or remote backends (e.g. S3, GCS, Azure Blob).The mlflow image is valuable for data science and MLOps teams who want to ensure reproducibility, observability, and deployment flexibility in their ML workflows.

How to use this image

The mlflow image can be used to track experiments, serve models, or run the tracking server.

Run the MLflow tracking server:

‍docker run -d -p 5000:5000 \  -v $(pwd)/mlruns:/mlflow/mlruns \  -e MLFLOW_BACKEND_STORE_URI=sqlite:///mlflow.db \  -e MLFLOW_DEFAULT_ARTIFACT_ROOT=/mlflow/mlruns \  mlflow/mlflow server \  --backend-store-uri sqlite:///mlflow.db \  --default-artifact-root /mlflow/mlruns

Serve a trained model:

docker run -p 1234:1234 \
  -v $(pwd)/model:/opt/ml/model \
  mlflow/mlflow models serve -m /opt/ml/model -p 1234

Use interactively to run MLflow CLI:

<code>Editdocker run -it mlflow/mlflow mlflow --help</code>

Logging and ports:

‍By default, the tracking server listens on port 5000, and the model server listens on port 1234. Logs are written to stdout.

Image variants

The mlflow image is published as mlflow/mlflow on Docker Hub and includes a single variant structure:

mlflow/mlflow:latest

Contains the latest released version of the MLflow CLI and server.Suitable for testing or development use.•

mlflow/mlflow:<version>

Tagged by MLflow version (e.g. 2.11.1, 2.9.0, etc.).Recommended for production use to ensure reproducibility and stability.There are no Alpine or OS-specific variants. The image is based on Python with common ML dependencies pre-installed.

Interested in base images that start and stay clean?

Oops! Something went wrong while submitting the form.