A Multichannel Biomedical Named Entity Recognition Model Based on Multitask Learning and Contextualized Word Representations

Table 1

Experimental parameter settings.

Hyperparameter	Value
Word dim	200
Char dim	30
ELMo dim	1024
GRU dim	100
Head	8
	1
Dropout rate	0.5
Initial learning rate	0.001
Optimizer	Adam
Batch size	32
Labeling schema	BIO