Yes. It’s very important. Think of it as quantifying whether a model understood the objective. This is, as you’d expect, especially important for auto-regressive models.