Connector: registry
Trains and promotes the same call against the local and mlflow registries and shows that versions, state and predictions are identical.
pip install pecca mlflow requests
jupyter lab examples/notebooks/connector_registry.ipynb
PECCA_SKIP_DOCKER=1 (as in CI) the Docker cells are skipped and the notebook still runs end to end. Outputs are committed, so GitHub renders the results without running anything.