Table 4.

Official results for the submitted runs along with the organizer’s baseline and an SVM model

Model	Run	RNN size	Shortcut connection	Precision	Recall	F1-score
Baseline	—	—	—	0.6122	0.6435	0.6274
SVM	—	—	—	0.5850	0.7869	0.6711
HBGRU	1	100	No	0.6136	0.7670	0.6818
	2	150	No	0.5944	0.8139	0.6871
	3	150	Yes	0.6289	0.7656	0.6906

The hyper-parameters not mentioned remain unchanged.