Title《Convolutional Sequence to Sequence Learning》 Abstract We introduce an architecture based entirely on convolutional

Title《Pay Less Attention With Lightweight and Dynamic Convolutions](https://arxiv.org/abs/1901.10430)》 Abstract self-att

Title《MUSE: PARALLEL MULTI-SCALE ATTENTION FOR SEQUENCE TO SEQUENCE LEARNING》 AbstractQ:attention mechanism alone suffer

Title:《Very Deep Transformers for Neural Machine Translation》 Q：how to decrease the variance of the output layer in orde

title:《Understanding Back-Translation at Scale》Q:how to improve neural machine translation with monolingual data? S:augm

《Task-Oriented Dialogue as Dataflow Synthesis》一、对话系统的问题和挑战 短句包括多个指示。 人类语言具有长尾性，无法使用高频意图覆盖 [^长尾性：系统中的个体相差悬殊，缺乏优选]: 多轮对