Integration · MLflow
MLflow Integration for the Model Registry
Retyping model metadata from MLflow by hand costs time and invites mistakes. With the MLflow integration, SimpleAct reads registered model versions and training metrics directly from your MLflow tracking server and imports them into the Model Registry on request – as a basis for Annex IV documentation.
Available in the product
SimpleAct connects to your MLflow tracking server, shows registered models with version, stage, and training metrics, and imports selected entries directly into the Model Registry – editable before saving.
- Browse registered models and versions from MLflow
- Auto-fill training metrics instead of retyping them by hand
- On-demand import – no automatic background sync
What the integration does
From MLflow straight into the Annex IV dossier
The integration reads what already exists in MLflow and makes it usable for compliance documentation – without SimpleAct running its own computation or duplicating data.
- Browse registered models
- SimpleAct lists the models registered in your MLflow Model Registry with their current version and stage (Staging, Production, Archived).
- Pull in training metrics
- Metrics from the associated MLflow run are auto-filled on import – you can still adjust them before saving.
- Traceable provenance
- Imported entries stay linked to the original MLflow model reference and run ID – traceable for audits.
Workflow
How to set up the MLflow integration
One-time connection, then sync manually whenever you need to.
Enter your tracking URI
Connect SimpleAct to your MLflow tracking server via its tracking URI (optionally with an access token for secured servers).
Browse and import models
Browse the MLflow models registered in the Model Registry, pick a model, and review the auto-filled details.
Complete before saving
Add model type, status, and the fairness checklist – only then is the entry saved in the Model Registry.
What teams can do with the MLflow integration
Built for existing MLflow setups – not a second model database running in parallel.
- Browse registered models and versions from MLflow
- Auto-fill training metrics
- Import targeted per model version, no automatic background sync
- Provenance (model reference, run ID) stays traceable
- Connect with or without an access token, including self-hosted MLflow servers
- Enterprise feature, configurable per tenant
Frequently asked questions about the MLflow integration
Does SimpleAct sync automatically in the background?
No. Import is deliberately on-demand: you browse the registered models and decide per version whether and when to import it.
Is training data or raw data transferred?
No. SimpleAct only reads model metadata and metrics from the MLflow API – no training or raw data.
Does this work with a self-hosted MLflow server?
Yes. The integration connects via the MLflow REST API and works with both self-hosted and managed tracking servers.
Model documentation that doesn't need to be maintained twice
With SimpleAct's MLflow integration, you pull model versions and metrics directly from your existing setup – as a basis for Annex IV documentation and audits.