Evaluating our model with create-evaluation