Table 3.

Accuracy (%) of RQE methods when trained using the training sets of SNLI, MultiNLI, Quora and Clinical-QE, and tested on our test set of 850 consumer health questions

Methods	Training Datasets
	SNLI	MultiNLI	Quora	Clinical-QE
Neural Network (NN)	48.94	54.59	52.35	48.71
NN + GloVe embeddings	49.41	54.82	52.82	57.18
Logistic Regression + Features	67.05	64.94	52.11	73.18

The best score are in bold