Table 1.

Ratios of the efficiencies obtained with samples using alternative MC generators, relative to those in the nominal Powhegbox + Pythia sample used during training of the algorithm

		70% OP		85% OP
		DL1d	GN2	DL1d	GN2
Powhegbox + Herwig	b-jets	0.984	0.984	0.989	0.990
	c-jets	0.951	0.904	0.977	0.983
	light-jets	1.003	1.000	1.015	1.011
Sherpa	b-jets	0.996	0.995	0.992	0.992
	c -jets	0.923	0.938	0.947	0.931
	light-jets	1.039	1.013	1.077	1.042

Statistical uncertainties from evaluating the same algorithm on different samples are negligible and thus not shown.