Convolutional Neural Networks for sentence Classification Model architecture Hyper-parameters Settings pretrain embedding : glove embedding size : 300 batch size : 100 filter size : [3,4,5] filter num : 100 dropout : 0.5 lr : 0.001