cws_lstm tensorflow bi-lstm for Chinese sentence segment files train.py training predict.py pridiction running train: python train.py predict: python predict.py result as following