catalog and pipeline_ml_factory have inconsistent artifact_path

[Galileo-Galilei]

log_model_kwargs are passed to the pipeline_ml_factory function:

kedro-mlflow-tutorial/src/kedro_mlflow_tutorial/pipeline_registry.py

Lines 30 to 38 in 67941e0

	log_model_kwargs=dict(
	artifact_path="kedro_mlflow_tutorial", # does not work with kedro==0.10.0
	conda_env={
	"python": 3.7,
	"build_dependencies": ["pip"],
	"dependencies": [f"kedro_mlflow_tutorial=={PROJECT_VERSION}"],
	},
	signature="auto",
	),

But they are ignored by kedro-mlflow and we need to use the default values when retrieveing the model.

kedro-mlflow-tutorial/conf/base/catalog.yml

Line 64 in 67941e0

artifact_path: model # it should be 'kedro_mlflow_tutorial', but there is a bug in kedro-mlflow==0.10.0

Once the issue is solved in kedro-mlflow, we should update it here.

[Galileo-Galilei]

Closed by 9e342f4