Research Article
Research on Named Entity Recognition Method of Metro On-Board Equipment Based on Multiheaded Self-Attention Mechanism and CNN-BiLSTM-CRF
| Parameter | Value |
| Batch_size | 64 | Learning rate | 8e − 4 | Hidden layer dimension | 128 | Epoch | 25 | LSTM_dim | 128 | cnn_size | 64 | attention_heads | 8 | kernel_size | (3, 4, 5) | Activation function | Relu | Max_seq_len | 128 | Dropout | 0.5 | Loss function | Cross entropy |
|
|