It looks to me like the plots meant to be generated by nn_tc_ever_v_never and structural_prior_min_freq are purported to only look at target ranks for the generative model, not others (pubchem/addcarbon etc), so I've incorporated those filters here.
The change in plots for PED dataset (enum factor 100) are:
It looks to me like the plots meant to be generated by
nn_tc_ever_v_neverandstructural_prior_min_freqare purported to only look at target ranks for the generative model, not others (pubchem/addcarbon etc), so I've incorporated those filters here.The change in plots for
PEDdataset (enum factor100) are:Before
After