# MLflow

MLflow is an open-source platform for managing the machine learning lifecycle, enabling teams to track experiments, evaluate and monitor LLM and agent applications, and deploy models to production. It is open source and self-hosted, installable on Docker Hub and Kubernetes (Helm).

Canonical page: https://ourlocalhost.com/app/4714/mlflow

## Description

MLflow is an open-source platform for managing the machine learning lifecycle, enabling teams to track experiments, evaluate and monitor LLM and agent applications, and deploy models to production.

## Platform availability

Available on 1 of 10 platforms: Helm.

## Health score

78/100 (Good), based on 4 of 4 factors.
- Maintained: 100 (last commit 4 days ago)
- Popular: 89 (28,858 GitHub stars)
- Easy to install: 25 (1 install platform, install notes)
- Light to run: 70 (1 GB minimum RAM)

## Links

- GitHub: https://github.com/mlflow/mlflow
- Docker Hub: https://hub.docker.com/r/ubuntu/mlflow
- Kubernetes (Helm): https://artifacthub.io/packages/helm/community-charts/mlflow
- Official website: https://mlflow.org

## Common questions

### Why does mlflow server not start after installing it?

It requires Python 3.10 or newer. Without a supported Python version, the server will not start.

### Why are logged runs or traces not visible in the MLflow UI?

The client tracking URI and the server backend-store URI must point to the same store. If they differ, the data may be saved but not visible in the UI.

### What port does the default local MLflow server use?

After starting the server with uvx mlflow server, the UI is available on localhost port 5000.

---
Source: Our Local Host (https://ourlocalhost.com/app/4714/mlflow), the self-hosted and open-source app directory.
