Using MLflow models in DSS

Scoring

Scoring recipe is available for MLflow saved models.

Two modes are available for scoring with an MLflow Saved Model:

1. Direct output mode where DSS outputs directly what the MLflow model outputs.

2. Restructure mode where DSS tries to interpret the model output.

Evaluation

You can evaluate versions of MLflow Saved Models on any valid DSS dataset. Related functionalities are available:

• Performance analysis

• Model Evaluation Store

• Model Comparison