VDCNN

https://arxiv.org/abs/1606.01781

$ python train.py train --batch-size {} --class-n {}