Table 2.

Hyperparameter settings.

Parameter	Value	Parameter	Value
Dropout rate	0.1	Filter num (DPCNNa)	256
Learning rate	1e-5	Filter size (DPCNN)	3
Max sequence length	56	Block size (DPCNN)	2
Optimizer	AdamW	Hidden size (LSTMb)	128
Training steps	30,000	Bidirectional (LSTM)	True

^aDPCNN: deep pyramid convolutional neural network.

^bLSTM: long short-term memory.