The inverse of the variance of the score function.

A sandwich-type estimator of the variance, $A B^{-1} A$, where $A$ is the expected Hessian and $B$ is the variance of the score.

The trace of the Fisher information matrix divided by $n$.

The difference between the observed and expected information.