Gotcha. That makes more sense. We ran the model to train the model -> “self-improving”.