Leveraging Contextual Sentences for Text Classification by Using a Neural Attention Model

Table 2

Experiment values and choices of some hyperparameters.

Hyperparameter	Choice	Experiment values
Learning rate	0.01	0.1, 0.01, 0.001
LSTM output dimension	100	50, 100, 150
LSTM direction	Bidir	Unidir, bidir
LSTM pooling	Last	Mean, last
CNN filter numbers	100	100, 300, 500, 700
CNN filter height	3	1, 2, 3, 4, 5
Word vector dimension	200	100, 200, 300