Table 9.

The results of different pre-trained models as well as Tavakoli et al.⁵⁰ on the test-B dataset.

Method	Lymp			Neut			Acc (%)
	P (%)	S (%)	F1 (%)	P (%)	S (%)	F1 (%)
ResNet1844	21	94	34	100	2	3	8
ResNet3444	24	94	38	100	27	43	32
ResNet5044	21	95	35	100	2	4	8
ResNext5045	24	90	38	100	3	5	9
MnasNet148	20	88	33	100	0	0	6
MobileNet-V247	72	50	59	100	0	0	4
DenseNet12146	43	64	51	100	8	15	12
ShuffleNet-V249	42	75	54	100	0	0	5
VGG1643	96	89	93	100	65	79	66
Tavakoli et al.50	94	54	69	100	92	96	90