A Hybrid Model Based on Deep Convolutional Network for Medical Named Entity Recognition

Table 4

Detailed parameters of the trained model.

Parameter variables	Values
Transformer layers	12
Hidden layers	768
Batch_size	8
Clip	0.5
Dropout_rate	0.5
Learning_rate	1e-5
Max_seq_length	202
Train_epochs	10
Warmup_proportion	0.1
lstm_size	128