Table 3 How good are the generated models? Description of the distributions of the TM-score quality. $\overline{fpt}$_<and $\overline{fpt}$_> are the relative frequencies of models per target with a TM-score below and above Arby default, respectively, $\overline{qir}$ is the improvement in TM-score when choosing models randomly. $\overline{fb}$ is the relative frequency of targets for which a better model exists, $\overline{qib}$ is the best theoretically possible improvement for the given ensemble of models.

$\overline{fpt}$ _<

$\overline{fpt}$ _>

$\overline{qir}$

$\overline{fb}$

$\overline{qib}$