Check test point is usually for checking whether the default value of your model makes sense. Usually if any number is NaN your model would immediately have problem.
The value beside each variable is the log probability, conditioned on the default value (i.e., model.test_point)