[Retracted] Parameter Optimization of Educational Network Ecosystem Based on BERT Deep Learning Model

Table 2

Experimental hyperparameter settings.

Parameter	MR	IMDB	SST-5(Sentence-level)	Tan Songbo hotel review
Convolution kernel width	2,1,4	3,2,5	1,4	2,3,5
Convolution kernel number	513	259	246	289
Number of hidden units	257	126	124	147
Batch size	59	38	61	31
Dropout	0.54	0.41	0.52	0.29