Table 2.

MetaVelvet-DL and -SL prediction accuracies on the CAMI low-complexity dataset

	MetaVelvet-DL	MetaVelvet-SL
Accuracy	55.6%	46.1%
Class 1 Sensitivity	35.9%	26.2%
Class 1 Specificity	82.9%	87.7%
Class 1 Balanced Accuracy	59.4%	56.9%
Class 2 Sensitivity	64.5%	54.5%
Class 2 Specificity	75.5%	63.6%
Class 2 Balanced Accuracy	70.0%	59.1%
Class 3 Sensitivity	60.1%	51.2%
Class 3 Specificity	70.1%	63.1%
Class 3 Balanced Accuracy	65.3%	57.2%