Table 2 Combinations of hyperparameters.

From: Domain adaptation using transformer models for automated detection of exterior cladding materials in street view images

Case

Hyperparameters

Batch size

Learning rate

Weight decay

Momentum

Iteration

1

1

0.00025

0.0001

0.7

3000

2

1

0.00025

0.0001

0.9

3000

3

1

0.00025

0.0005

0.7

3000

4

1

0.00025

0.0005

0.9

3000

5

1

0.001

0.0001

0.7

3000

6

1

0.001

0.0001

0.9

3000

7

1

0.001

0.0005

0.7

3000

8

1

0.001

0.0005

0.9

3000

9

2

0.00025

0.0001

0.7

3000

10

2

0.00025

0.0001

0.9

3000

11

2

0.00025

0.0005

0.7

3000

12

2

0.00025

0.0005

0.9

3000

13

2

0.001

0.0001

0.7

3000

14

2

0.001

0.0001

0.9

3000

15

2

0.001

0.0005

0.7

3000

16

2

0.001

0.0005

0.9

3000

17

1

0.00025

0.0001

0.7

5000

18

1

0.00025

0.0001

0.9

5000

19

1

0.00025

0.0005

0.7

5000

20

1

0.00025

0.0005

0.9

5000

21

1

0.001

0.0001

0.7

5000

22

1

0.001

0.0001

0.9

5000

23

1

0.001

0.0005

0.7

5000

24

1

0.001

0.0005

0.9

5000

25

2

0.00025

0.0001

0.7

5000

26

2

0.00025

0.0001

0.9

5000

27

2

0.00025

0.0005

0.7

5000

28

2

0.00025

0.0005

0.9

5000

29

2

0.001

0.0001

0.7

5000

30

2

0.001

0.0001

0.9

5000

31

2

0.001

0.0005

0.7

5000

32

2

0.001

0.0005

0.9

5000