mlflow
mlflow — ɳSelf documentation.
MLflow Plugin
ML experiment tracking, model registry, and artifact storage. Free: MIT licensed.
Install
nself plugin install mlflowWhat It Does
Runs an MLflow tracking server backed by your Postgres database. Track experiments, compare model runs, log parameters and metrics, and manage the model registry. Integrates with Python ML workflows via the MLflow client.
Configuration
| Env Var | Default | Description |
|---|---|---|
MLFLOW_PORT | 5000 | MLflow server port |
MLFLOW_ARTIFACT_ROOT | /mlflow/artifacts | Artifact storage path |
Ports
| Port | Purpose |
|---|---|
| 5000 | MLflow UI and REST API |
Database Tables
0 plugin-managed tables: MLflow manages its own schema in Postgres directly via SQLAlchemy.
Nginx Routes
| Route | Target |
|---|---|
mlflow.{base_domain} | MLflow UI |
Dependencies
Requires Postgres (always running in ɳSelf). Optionally uses MinIO for artifact storage, set MLFLOW_ARTIFACT_ROOT=s3://... and configure MinIO credentials.