Table 1. Parameter setting.

Hyperparameter	Value
Learning rate	0.001
Optimizer	AdamW
Regularization	L2 regularization, coefficient 0.01
Dropout	0.2
Batch size	32
Epochs	100